chore(release): prepare v0.19.6

feat(anki): add media timing review before card creation (#203 )
fix(tsukihime): filter releases by subtitle language (#235 )
2026-09-04 11:54:28 -07:00 · 2026-09-04 01:59:39 -07:00 · 2026-09-04 01:40:56 -07:00 · 2026-09-02 20:29:59 -07:00 · 2026-09-02 20:16:08 -07:00 · 2026-09-01 23:11:38 -07:00
195 changed files with 16186 additions and 1411 deletions
@@ -0,0 +1,34 @@
+---
+name: subminer-release
+description: Prepare, cut, publish, or repair SubMiner stable and prerelease releases. Use for hands-on release work; do not use for general release questions.
+---
+
+# SubMiner release
+
+Carry out the requested release phase using the repository's current release process.
+
+## Source of truth
+
+Read `docs/RELEASING.md` completely before changing files or release state. Treat it as canonical. Read `changes/README.md` when the work touches change fragments or generated release notes.
+
+Do not copy release commands or policy into this skill. If this skill disagrees with the release guide, follow the guide and reconcile the skill before handoff.
+
+## Workflow
+
+1. Identify whether the request is for a stable release, prerelease, release preparation, publication, or repair.
+2. Inspect the current branch, worktree status, package version, pending change fragments, relevant tags, and latest CI state before making changes.
+3. Follow the matching procedure in `docs/RELEASING.md` in order. Review generated changelog and release-note Markdown before it can be committed or published.
+4. Run every required gate for the requested release phase. Do not treat a cheaper test lane as a substitute for the documented release gate.
+5. Before a stable tag, confirm the package and tag versions match and no pending `changes/*.md` fragments remain. Preserve fragments for prereleases as documented.
+6. Report the resulting version, completed checks, local commit and tag state, remote publication state, skipped platform checks, and any remaining manual work.
+
+## Authorization boundaries
+
+- A request to prepare a release stops before commit, tag, push, or remote publication unless the user also authorizes those actions.
+- A clear request to cut or publish a release includes the documented commit, tag, and push steps. Ask before the first remote mutation when the wording is ambiguous.
+- Do not edit an existing GitHub release, publish to the AUR, change secrets, or alter signing configuration unless the user explicitly requests that operation.
+- Do not switch branches without consent.
+
+## Stop conditions
+
+Stop and report the blocker when required CI or a release gate fails, authentication is missing, versions disagree, required artifacts are absent, or the worktree contains unexpected changes that overlap the release. Do not tag or publish a partially verified release.
@@ -32,9 +32,11 @@ jobs:
      - name: Guard stable docs tag shape
        id: tag_guard
        if: github.ref_type == 'tag'
+        env:
+          TAG_NAME: ${{ github.ref_name }}
        run: |
-          if [[ ! "${{ github.ref_name }}" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
-            echo "::notice::Skipping non-stable docs tag ${{ github.ref_name }}"
+          if [[ ! "$TAG_NAME" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+            echo "::notice::Skipping non-stable docs tag $TAG_NAME"
            echo "stable_tag=false" >> "$GITHUB_OUTPUT"
            exit 0
          fi
@@ -297,15 +297,22 @@ jobs:
        run: echo "VERSION=${GITHUB_REF#refs/tags/}" >> "$GITHUB_OUTPUT"

      - name: Verify committed prerelease notes
+        env:
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          if [ ! -s release/prerelease-notes.md ]; then
            echo "::error::release/prerelease-notes.md is missing or empty. Run 'bun run changelog:prerelease-notes --version <version>' locally and commit the file before tagging."
            exit 1
          fi
+          if ! bun run changelog:check-prerelease-notes --version "$RELEASE_VERSION"; then
+            echo "::error::release/prerelease-notes.md was not generated for $RELEASE_VERSION. Rerun 'bun run changelog:prerelease-notes --version $RELEASE_VERSION' locally, commit, and retag."
+            exit 1
+          fi

      - name: Publish Prerelease
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          set -euo pipefail

@@ -327,27 +334,27 @@ jobs:
            exit 1
          fi

-          if gh release view "${{ steps.version.outputs.VERSION }}" >/dev/null 2>&1; then
-            gh release edit "${{ steps.version.outputs.VERSION }}" \
+          if gh release view "$RELEASE_VERSION" >/dev/null 2>&1; then
+            gh release edit "$RELEASE_VERSION" \
              --draft \
              --prerelease \
-              --title "${{ steps.version.outputs.VERSION }}" \
+              --title "$RELEASE_VERSION" \
              --notes-file release/prerelease-notes.md
          else
-            gh release create "${{ steps.version.outputs.VERSION }}" \
+            gh release create "$RELEASE_VERSION" \
              --draft \
              --latest=false \
              --prerelease \
-              --title "${{ steps.version.outputs.VERSION }}" \
+              --title "$RELEASE_VERSION" \
              --notes-file release/prerelease-notes.md
          fi

          for asset in "${artifacts[@]}"; do
-            gh release upload "${{ steps.version.outputs.VERSION }}" "$asset" --clobber
+            gh release upload "$RELEASE_VERSION" "$asset" --clobber
          done

-          gh release edit "${{ steps.version.outputs.VERSION }}" \
+          gh release edit "$RELEASE_VERSION" \
            --draft=false \
            --prerelease \
-            --title "${{ steps.version.outputs.VERSION }}" \
+            --title "$RELEASE_VERSION" \
            --notes-file release/prerelease-notes.md
@@ -296,33 +296,40 @@ jobs:
        run: echo "VERSION=${GITHUB_REF#refs/tags/}" >> "$GITHUB_OUTPUT"

      - name: Guard against pending changelog fragments
+        env:
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          if find changes -maxdepth 1 -name '*.md' -not -name README.md -print -quit | grep -q .; then
-            echo "::error::Pending changelog fragments detected. Run 'bun run changelog:build --version ${{ steps.version.outputs.VERSION }}' locally and commit the polished CHANGELOG.md before tagging. CI no longer auto-builds the changelog because the polish step requires the local 'claude' CLI."
+            echo "::error::Pending changelog fragments detected. Run 'bun run changelog:build --version $RELEASE_VERSION' locally and commit the polished CHANGELOG.md before tagging. CI no longer auto-builds the changelog because the polish step requires the local 'claude' CLI."
            exit 1
          fi

      - name: Verify changelog is ready for tagged release
-        run: bun run changelog:check --version "${{ steps.version.outputs.VERSION }}"
+        env:
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
+        run: bun run changelog:check --version "$RELEASE_VERSION"

      - name: Generate release notes from changelog
-        run: bun run changelog:release-notes --version "${{ steps.version.outputs.VERSION }}"
+        env:
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
+        run: bun run changelog:release-notes --version "$RELEASE_VERSION"

      - name: Publish Release
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          set -euo pipefail

-          if gh release view "${{ steps.version.outputs.VERSION }}" >/dev/null 2>&1; then
+          if gh release view "$RELEASE_VERSION" >/dev/null 2>&1; then
            # Do not pass the prerelease flag here; gh defaults to a normal release.
-            gh release edit "${{ steps.version.outputs.VERSION }}" \
+            gh release edit "$RELEASE_VERSION" \
              --draft=false \
-              --title "${{ steps.version.outputs.VERSION }}" \
+              --title "$RELEASE_VERSION" \
              --notes-file release/release-notes.md
          else
-            gh release create "${{ steps.version.outputs.VERSION }}" \
-              --title "${{ steps.version.outputs.VERSION }}" \
+            gh release create "$RELEASE_VERSION" \
+              --title "$RELEASE_VERSION" \
              --notes-file release/release-notes.md
          fi

@@ -345,7 +352,7 @@ jobs:
          fi

          for asset in "${artifacts[@]}"; do
-            gh release upload "${{ steps.version.outputs.VERSION }}" "$asset" --clobber
+            gh release upload "$RELEASE_VERSION" "$asset" --clobber
          done

  aur-publish:
@@ -421,9 +428,10 @@ jobs:
        if: steps.aur_prereqs.outputs.skip != 'true' && steps.aur_ssh.outputs.skip != 'true' && steps.aur_clone.outputs.skip != 'true'
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          set -euo pipefail
-          version="${{ steps.version.outputs.VERSION }}"
+          version="$RELEASE_VERSION"
          install -dm755 .tmp/aur-release-assets
          gh release download "$version" \
            --dir .tmp/aur-release-assets \
@@ -433,15 +441,17 @@ jobs:

      - name: Update AUR packaging metadata
        if: steps.aur_prereqs.outputs.skip != 'true' && steps.aur_ssh.outputs.skip != 'true' && steps.aur_clone.outputs.skip != 'true'
+        env:
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          set -euo pipefail
-          version_no_v="${{ steps.version.outputs.VERSION }}"
+          version_no_v="$RELEASE_VERSION"
          version_no_v="${version_no_v#v}"
          cp packaging/aur/subminer-bin/PKGBUILD aur-subminer-bin/PKGBUILD
          cp packaging/aur/subminer-bin/.SRCINFO aur-subminer-bin/.SRCINFO
          bash scripts/update-aur-package.sh \
            --pkg-dir aur-subminer-bin \
-            --version "${{ steps.version.outputs.VERSION }}" \
+            --version "$RELEASE_VERSION" \
            --appimage ".tmp/aur-release-assets/SubMiner-${version_no_v}.AppImage" \
            --wrapper ".tmp/aur-release-assets/subminer" \
            --assets ".tmp/aur-release-assets/subminer-assets.tar.gz"
@@ -451,6 +461,7 @@ jobs:
        working-directory: aur-subminer-bin
        env:
          GIT_SSH_COMMAND: ssh -i ~/.ssh/aur -o IdentitiesOnly=yes
+          RELEASE_VERSION: ${{ steps.version.outputs.VERSION }}
        run: |
          set -euo pipefail
          if git diff --quiet -- PKGBUILD .SRCINFO; then
@@ -460,7 +471,7 @@ jobs:
          git config user.name "github-actions[bot]"
          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
          git add PKGBUILD .SRCINFO
-          git commit -m "Update to ${{ steps.version.outputs.VERSION }}"
+          git commit -m "Update to $RELEASE_VERSION"

          attempts=3
          for attempt in $(seq 1 "$attempts"); do
@@ -49,6 +49,7 @@ tests/*
 !.agents/skills/
 .agents/skills/*
 !.agents/skills/subminer-change-verification/
+!.agents/skills/subminer-release/
 !.agents/skills/subminer-scrum-master/
 .agents/skills/subminer-change-verification/*
 !.agents/skills/subminer-change-verification/SKILL.md
@@ -56,6 +57,8 @@ tests/*
 .agents/skills/subminer-change-verification/scripts/*
 !.agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh
 !.agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh
+.agents/skills/subminer-release/*
+!.agents/skills/subminer-release/SKILL.md
 .agents/skills/subminer-scrum-master/*
 !.agents/skills/subminer-scrum-master/SKILL.md
 favicon.png
@@ -1,5 +1,92 @@
 # Changelog

+## v0.19.6 (2026-09-04)
+
+### Added
+
+- **Card Timing Review**:
+  - Optional pre-generation timing review for word, sentence, and audio cards, with a speech-weighted waveform that flattens background noise so dialogue edges stand out clearly.
+  - The clip end automatically snaps back to where the line's dialogue actually ends once the waveform loads, with drag and keyboard adjustments available.
+  - Audio preview includes a sweeping playhead that plays the clip to its true end, even on high-latency outputs like Bluetooth headphones.
+  - Previous and next subtitle lines can be pulled onto the card with `P`/`N` (or the Prev/Next steppers) and removed with Shift; the sentence preview and waveform markers update automatically.
+  - Cancelling lets you keep a card without media, and the review can be toggled on or off for the session.
+- **Senren Field Grouping**:
+  - Enable `ankiConnect.isSenren` to merge duplicate mined cards using Senren's scene-switching markup, grouping sentence, furigana, audio, picture, and misc-info fields.
+  - Supports the same auto/manual/disabled modes as Kiku, including the manual merge modal; only one of Senren or Kiku can be enabled at a time.
+
+### Changed
+
+- **Remote Stream Mining Performance**: Mining a card from a remote stream (Jellyfin and other HTTP sources) now downloads the clip window once and reuses it for the timing review waveform, audio preview, audio extraction, and screenshot, instead of re-fetching the stream at each step; the temporary file is cleaned up after ten minutes of inactivity or on exit.
+- **TsukiHime Release Filtering**: The TsukiHime modal's Japanese and secondary-language tabs now filter the release list by the subtitle languages each release actually carries, and report when no release has subtitles for the active tab.
+
+### Fixed
+
+- **Subtitle & Mining Accuracy**:
+  - Broadcast-style captions that split one sentence across two on-screen rows (e.g. Crunchyroll Japanese subs) now merge into a single line for the sidebar and mined cards, while separate speakers, sound effects, and labeled turns still stay on their own lines.
+  - Mining from the overlay no longer pulls in a lingering row from the previous caption; the mined sentence and clip timing now match what's actually on screen.
+  - Multi-line copy and mining now select lines backward in timeline order after seeking, instead of in playback encounter order.
+  - Copying a subtitle, mining a sentence, or recording immersion stats no longer includes the separate furigana line that broadcast ASS captions place above a word.
+- **Card Update Notifications**: Dismissed lingering overlay card-update progress when notification settings switch to OSD before an update finishes.
+- **Overlay Stability on Hyprland**: Opening a modal window (timing review, Jimaku, session help, and others) while mpv is fullscreen no longer causes the overlay to flicker while the modal loads; the overlay now stays on screen untouched until the modal is ready.
+- **Jellyfin Subtitle Sync**: Jellyfin subtitle files now load with zero mpv delay instead of inferring and saving an offset from Japanese and English cue timelines.
+- **Secondary Subtitle Visibility**: Native mpv secondary subtitles stay hidden when switching secondary subtitle tracks during playback.
+
+## v0.19.5 (2026-08-30)
+
+### Fixed
+
+- **Anki Card Update Progress**: The card-update spinner now stays visible until audio and image updates finish, instead of disappearing early.
+- **Anki Word-Card Fields**: Word-card enrichment now writes sentence text and audio to the fields configured in AnkiConnect, while the dedicated sentence-card and audio-card actions keep their existing compatible field names.
+- **Overlapping Subtitles**:
+  - Subtitle lines that start while another line is still on screen now appear alongside it, instead of staying hidden until a track switch or seek.
+  - Subtitles shown at the same time now stack by their authored screen position, with top signs and song lines above bottom dialogue.
+  - Half-size ASS furigana is no longer shown as if it were a dialogue line.
+- **YouTube Auto Captions**:
+  - Auto-generated captions now follow their intended timing and two-row roll-up layout.
+  - Long speech is paged instead of covering the video with a wall of text.
+  - Explicitly timed sound cues like `[音楽]` no longer cover later dialogue.
+
+## v0.19.4 (2026-08-25)
+
+### Added
+- **Library Merge & Move**: Duplicate library cards for the same show can now be combined. Select cards in the library grid and use "Merge Selected" to pick which entry to keep and move every episode onto it, preserving sessions, mined cards, and watch time. Episodes can also be reassigned individually via the "→" button, useful when a file lands under a stray title; manual assignments survive later filename parsing, Jellyfin refreshes, and season repair. Exact AniList title matches with compatible seasons now merge automatically, while fuzzy matches surface as dismissible "Possible duplicate" reviews instead of merging silently.
+- **Duplicate Line Cleanup Tool**: The Vocabulary tab's new "Duplicates" button scans a chosen time window (7 days through all time) for old karaoke/typeset duplicate-line bursts, shows what it found, and collapses each run to one line once confirmed; `subminer stats cleanup --duplicate-lines` does the same from the terminal, with `--dry-run` and `--lookback-days <n>` options. Watch time and lines-seen totals are left unchanged.
+
+### Changed
+- **Prerelease Release Notes**: Prerelease notes now open with a "Changes since" section listing only what changed versus the previous beta/RC of the same version, above the cumulative highlights, and CI rejects prerelease tags whose committed notes were generated for a different beta/RC.
+
+### Fixed
+- **Subtitle & Karaoke Duplication**:
+  - Karaoke and animated signs are reconstructed once from their authored text and shown only while actually sung, with original word spacing preserved, instead of flooding the overlay, subtitle sidebar, immersion history, mining, or stats with glyph fragments, per-frame color phases, and repeated animation events.
+  - Decorative layers (highlight sweeps, glow/shadow copies, symbol-font decoration, particle swarms, hidden or zero-scaled text) stay out of published text, while ordinary repeated dialogue, positioned signs, wrapped lyric rows, and multi-row CC-style blocks still display correctly.
+  - Embedded subtitle tracks on network-mounted (SMB/NFS) media are extracted and parsed again instead of falling back to live-text-only, restoring karaoke reconstruction, sidebar cues, and mining for releases that only ship subtitles inside the container.
+  - Secondary subtitles go through the same deduplication pipeline as primary subtitles and no longer clip display after about four lines.
+  - Event-heavy karaoke files that previously stalled subtitle loading for several seconds now parse in well under a second.
+- **Character Dictionary Reliability**:
+  - Generation, merged rebuilds, and imports no longer freeze the app on large dictionaries; snapshot I/O, archive building, and image/name lookup caches moved off the UI's critical path.
+  - Dictionaries are reused instead of regenerated when MeCab finds no name splits.
+  - Cached portraits restore correctly after the portrait index finishes loading post-tokenization.
+  - Desktop progress notifications on Linux AppImage installs update in place instead of flickering, fixing a bug where the AppImage's bundled libraries broke the system notification helper.
+- **Overlay Startup & Modals**:
+  - The macOS window-tracking helper targets macOS 12.0+ instead of requiring the build machine's exact macOS version, fixing crashes on older systems like Ventura that left the overlay stuck on "Overlay loading".
+  - mpv IPC connection attempts time out and retry, showing an actionable error if content still isn't ready after 30 seconds.
+  - Dedicated overlay modals are prewarmed on macOS and Windows so shortcuts open them promptly.
+  - On macOS, reused modals and the stats window open above fullscreen mpv on its current Space instead of jumping to another desktop.
+- **Wayland File Drop**: Fixed native Wayland drag-and-drop from file managers such as Thunar, so subtitle and video files dropped on the visible overlay are resolved and forwarded to mpv.
+- **Windows Mouse Lag**: Fixed system-wide mouse lag on Windows while SubMiner is running, caused by the overlay's global mouse hook for click-through forwarding and by the mpv window tracker blocking the app on repeated PowerShell lookups.
+- **Sentence Mining Audio & Clips**: Sentence-audio generation no longer times out on slow network-mounted media with many subtitle/font streams (bounded FFmpeg probing, two-minute extraction budget, clearer error reporting), and mined audio/animated AVIF clips now capture the subtitle line that was actually mined by snapshotting the clip range at lookup time instead of reading live mpv state later.
+- **Stats Performance & Reliability**: Immersion stats storage now sets its SQLite busy timeout before WAL setup, avoiding transient lock errors under concurrent writes. Deletes in the stats dashboard no longer freeze the UI, run proportional to what's deleted instead of rebuilding full lifetime summaries, retry safely if the delete worker crashes, and no longer rescan the whole library when deleting very common words; a new index also makes large session deletes drop from minutes to milliseconds. Library merges, video moves, and AniList reassignments got the same lifetime-summary fix.
+- **Vocabulary Stats Accuracy**: Vocabulary totals and charts now count all tracked vocabulary instead of only the first page, new-word history uses corrected daily rollups (fixing legacy timestamp and time-zone issues), summary cards refresh automatically after edits to the exclusion list, and rapid exclusion edits no longer race each other.
+- **Rofi MKV Thumbnails**: Fixed missing MKV thumbnails in the Linux rofi picker when system thumbnailer registrations only advertise legacy Matroska MIME aliases.
+
+<details>
+<summary>Internal changes</summary>
+
+### Internal
+- Docs Site Indexing: Excluded the `/main/` and `/v/<version>/` docs trees from search indexing (self-referential canonical, `noindex,follow`, matching `X-Robots-Tag`) so crawlers focus on current docs instead of ~30 archived copies of every page, and restored `<lastmod>` dates in the docs sitemap that were silently dropped by production builds.
+
+</details>
+
 ## v0.19.3 (2026-08-13)

 ### Added
@@ -38,7 +38,7 @@
    "app-builder-lib": "26.15.3",
    "brace-expansion": "5.0.9",
    "electron-builder-squirrel-windows": "26.15.3",
-    "fast-uri": "3.1.5",
+    "fast-uri": "3.1.6",
    "form-data": "4.0.6",
    "ip-address": "10.2.0",
    "js-yaml": "4.3.1",
@@ -406,7 +406,7 @@

    "fast-levenshtein": ["fast-levenshtein@2.0.6", "", {}, "sha512-DCXu6Ifhqcks7TZKY3Hxp3y6qphY5SJZmrWMDrKcERSOXWQdMhU9Ig/PYrzyw/ul9jOIyh0N4M0tbC5hodg8dw=="],

-    "fast-uri": ["fast-uri@3.1.5", "", {}, "sha512-gHwA1O9LDIcKunMKhObS/HimwtehO1nPUECKAu5TpKgaO19fcWEl4bliWe1jWxVFvIXztJjjQ4L8XQ1EU9f7Jw=="],
+    "fast-uri": ["fast-uri@3.1.6", "", {}, "sha512-7Ical1vFEMr0onbVzEDIreM22I4khW+fzyQPwvAFWBp1iwdshSZRsL4jjRvPG9JP1uiqMHRto+YU6R2/CzDz5Q=="],

    "fdir": ["fdir@6.5.0", "", { "peerDependencies": { "picomatch": "^3 || ^4" }, "optionalPeers": ["picomatch"] }, "sha512-tIbYtZbucOs0BRGqPJkshJUYdL+SDH7dVM8gjy+ERp3WAUjLEFJE+02kanyHtwjWOnwrKYBiwAmM0p4kLJAnXg=="],

@@ -42,13 +42,14 @@ How fragments turn into a release:

 - At release time, `bun run changelog:build` (and `bun run changelog:prerelease-notes`) pipes every pending fragment through `claude -p` to merge related items, drop noise, and rewrite into a clean user-facing release body. Write fragments as raw, informative notes — don't worry about polished prose, deduping across PRs, or line-by-line phrasing. The polish step handles all of that.
 - The polish step treats pending fragments as the final release outcome, not prerelease history. If a feature is added and then renamed or fixed before the stable cut, ship the final feature bullet instead of separate prerelease-only breaking/fix entries.
- GitHub release notes and prerelease notes use short top-level items with nested bullets for the change, user benefit, and any useful action note. The stable `CHANGELOG.md` can stay in compact single-line bullets.
+- `CHANGELOG.md`, GitHub release notes, and prerelease notes all use short top-level items with one nested bullet per distinct change, instead of packing a release's worth of detail into a single paragraph bullet. An item with only one thing to say stays inline on the top-level bullet. Release notes and prerelease notes additionally cover user benefit and any useful action note in their nested bullets.
 - `internal` fragments stay in `CHANGELOG.md` (inside a collapsed `<details>` block) but are dropped from the GitHub release notes entirely.
 - The polished `CHANGELOG.md` and `release/release-notes.md` are committed and reviewed before tagging — edit the Markdown by hand if Claude misses something.

 Prerelease notes:

 - prerelease tags like `v0.11.3-beta.1` and `v0.11.3-rc.1` reuse the current pending fragments to generate `release/prerelease-notes.md`
+- from the second prerelease of a base version onward, the notes also open with a `## Changes since <previous tag>` section generated from the fragment diff against the previous beta/RC tag; keep fragment edits meaningful. Editorial-only rewording is filtered out of that section, while genuinely changed behavior and deleted fragments (reverted changes) are reported
 - existing prerelease notes are a reviewed baseline; later prerelease runs should replace stale beta/RC wording with the current outcome instead of appending fix churn
 - prerelease note generation does not consume fragments and does not update `CHANGELOG.md` or `docs-site/changelog.md`
 - the final stable release is the point where `bun run changelog:build` consumes fragments into the stable changelog and release notes
@@ -1,5 +0,0 @@
-type: fixed
-area: subtitles
-
- Typeset ASS karaoke and animated signs no longer flood the primary overlay, subtitle sidebar, immersion history, or sentence mining with repeated glyph fragments or full-line color phases. Matching timed comments and full-line boundary events recover the complete authored line without merging ordinary repeated dialogue or separately positioned signs, and dialogue spoken while a song's animation is on screen is kept intact instead of being replaced by the lyric. Entrance and exit frames that run past the authored line timing still resolve to the clean line during lyric transitions, and dialogue spoken while a song's animation is on screen enters immersion and subtitle history without the fragment lines beside it.
- The secondary subtitle overlay drops layered duplicate lines from animated tracks, so a short stack of repeated words collapses to its distinct lines even when the full karaoke heuristic does not apply.
@@ -1,4 +0,0 @@
-type: fixed
-area: Anki media
-
- Fixed sentence-audio generation timing out on slow network-mounted MKV files with many subtitle and font-attachment streams. Selected audio tracks now use bounded FFmpeg probing and a two-minute extraction budget, and missing output reports a clear FFmpeg error instead of raw `ENOENT`.
@@ -1,5 +0,0 @@
-type: fixed
-area: character dictionary
-
- Reuse character dictionaries after MeCab completes without finding any name splits instead of regenerating character data and portraits on every launch.
- Restore inline character portraits when a cached portrait index finishes loading after subtitles have already been tokenized.
@@ -1,5 +0,0 @@
-type: fixed
-area: dictionary
-
- Character dictionary generation, merged rebuilds, and imports no longer freeze the app (and trigger the compositor's "application not responding" dialog) on large dictionaries; snapshot reads/writes, archive building, and the character image/name lookup caches now do their heavy work off the UI's critical path.
- Desktop progress notifications now update in place on Linux AppImage installs too: the AppImage's bundled libraries broke the system notify-send helper, which silently forced the flickering close-and-reopen notification fallback.
@@ -1,5 +0,0 @@
-type: internal
-area: docs
-
- Excluded the `/main/` and `/v/<version>/` docs trees from search indexing with a self-referential canonical, `noindex,follow`, and a matching `X-Robots-Tag` header, so crawlers spend their budget on the current docs instead of ~30 archived copies of every page.
- Restored `<lastmod>` dates in the docs sitemap, which were silently dropped because production builds render from an untracked release snapshot.
@@ -1,5 +0,0 @@
-type: fixed
-area: stats
-
- Typeset subtitles no longer flood the stats. Karaoke openings and animated signs are authored as one subtitle event per animation frame, and immersion tracking counted every frame, which was enough to put an OP lyric at the top of "Top Repeated Words" for good. Lines are now collapsed on the way in using the same rules the subtitle sidebar already applies: matching parsed timings record exactly the cues the sidebar shows, while shifted, changing, or unparsed sources use a strict fallback where identical, contiguous, sub-0.1s lines stop counting after a few frames. Ordinary repeated dialogue and rewatches are unaffected.
- Added a cleanup for stats already affected. The Vocabulary tab has a **Duplicates** button that scans a chosen window (7 days through all time), shows the bursts it found and the word and kanji counts they added, and collapses each run to one line once confirmed. `subminer stats cleanup --duplicate-lines` does the same from the terminal, with `--dry-run` and `--lookback-days <n>`. Only subtitle lines and the vocabulary counts they feed are touched; watch time and lines-seen totals are left as recorded.
@@ -1,5 +0,0 @@
-type: fixed
-area: overlay
-
- Dedicated overlay modals are prewarmed on macOS and Windows so shortcuts open them promptly on the first press. Windows now refreshes the hidden modal renderer between sessions to keep later modals interactive. On macOS, reused modals and the in-app stats window also open above fullscreen mpv on its current Space instead of appearing on another desktop or forcing a Space change.
- Updated subtitle ASS observation to mpv's current `sub-text/ass` property, removing its deprecation warning.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Fixed the overlay getting stuck on "Overlay loading" forever when startup stalls: mpv IPC connection attempts now time out and retry, switching sockets aborts obsolete attempts, and the plugin replaces its spinner with an actionable error if overlay content is still not ready after 30 seconds.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Fixed native Wayland drag-and-drop from file managers such as Thunar so subtitle and video files dropped on the visible overlay are resolved and forwarded to mpv.
@@ -1,4 +0,0 @@
-type: fixed
-area: subtitles
-
- Primary ASS subtitles now use the active parsed cue when it fully accounts for mpv's live text, preventing fill, border, blur, and shadow copies of the same full-span lyric from appearing repeatedly while preserving unmatched overlapping dialogue and signs.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Secondary subtitles now parse the selected ASS/SRT/VTT source with the primary subtitle deduplication pipeline, preventing layered animation text from appearing several times in the overlay, mined cards, and statistics. Live mpv text remains the fallback for unreadable tracks.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Fixed system-wide mouse lag on Windows while SubMiner is running: the overlay no longer installs Electron's global mouse hook for click-through forwarding, and the mpv window tracker no longer blocks the app on repeated PowerShell command-line lookups.
@@ -1,6 +0,0 @@
-type: added
-area: stats
-
- Library: duplicate cards for the same show can now be combined. Press "Select" above the library grid, tick the cards, and use "Merge Selected"; the dialog picks which entry to keep and moves every episode onto it. Sessions, mined cards, and watch time are preserved, the emptied entries disappear, and remembered title aliases keep future episodes on the merged card.
- Library: episodes can be reassigned to another library entry from the "→" button on an episode row, which is the fix when one file lands under a stray title (e.g. an episode name parsed as the series). Manual assignments now survive later filename parsing, Jellyfin refreshes, and season repair. Local episodes in the same directory reuse a uniquely corrected destination unless they parse to a title that already has its own library entry, while conflicting seasons or manual destinations are not forced together. Emptying an entry this way removes it and returns to the grid.
- Library: exact AniList title matches with compatible seasons fold duplicate cards automatically. Fuzzy same-AniList matches appear as dismissible "Possible duplicate" reviews instead of changing the library without confirmation; conflicting explicit seasons are left alone.
@@ -1,4 +0,0 @@
-type: fixed
-area: notifications
-
- Character dictionary progress notifications on Linux now update in place instead of flickering off and reappearing on every status change.
@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
- Mined audio and animated AVIF clips now capture the subtitle line that was actually mined. The clip range is snapshotted once at Yomitan lookup time (and reused for both audio and image), instead of each generator reading the live mpv subtitle when it starts — which clipped whatever line was on screen after slow audio extraction finished, producing too-short or misaligned AVIF clips.
@@ -1,4 +0,0 @@
-type: fixed
-area: launcher
-
- Fixed missing MKV thumbnails in the Linux rofi picker when system thumbnailer registrations only advertise legacy Matroska MIME aliases.
@@ -1,9 +0,0 @@
-type: fixed
-area: stats
-
- Stats deletes no longer freeze the stats dashboard: the delete worker module now resolves when running from source, so deletes actually run off the serving thread instead of silently falling back to it.
- Deletes now subtract their exact contribution from lifetime summaries instead of rebuilding them from retained sessions, making delete cost proportional to what is deleted and preserving lifetime totals older than the session retention window.
- If the delete worker crashes, the delete now retries on the current thread instead of failing.
- Library merges, video moves, AniList reassignments, and `subminer stats cleanup -l` also stopped rebuilding lifetime summaries from retained sessions; they now recompute from per-episode history, so those operations are faster and no longer erase lifetime totals older than the session retention window.
- Deleting content that contains very common words no longer rescans every occurrence of those words across the whole library; first/last-seen dates are refreshed with index seeks instead.
- Session deletes on large databases dropped from minutes to milliseconds: an index on the subtitle-line event reference now prevents each deleted session event from scanning the whole subtitle-line table for foreign-key enforcement.
@@ -1,8 +0,0 @@
-type: fixed
-area: stats
-
- Fixed Vocabulary totals and charts counting only the first browsing page instead of all tracked vocabulary, without delaying the rest of the page.
- New-word history now uses permanent daily lexical rollups that apply the same vocabulary filters as the totals and normalize legacy second/millisecond timestamps; versioned background rebuilds repair existing history across legacy rollup-state schemas without dropping playback writes or clearing watch-time, activity, efficiency, and library charts.
- Calendar-day chart labels now preserve the recorded local date in time zones west of UTC.
- Vocabulary summary cards and charts refresh automatically after the word exclusion list changes, and failed or unfinished loads use bounded retries before showing an inline error with a Retry control.
- Rapid exclusion edits no longer race each other; writes are sent in order so a slower earlier save cannot overwrite a newer list.
@@ -523,7 +523,7 @@
  // ==========================================
  // AnkiConnect Integration
  // Automatic Anki updates and media generation options.
-  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.
+  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume/reviewTiming, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, isSenren.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.
  // Shared AI provider transport settings are read from top-level ai and typically require restart.
  // Most other AnkiConnect settings still require restart.
  // ==========================================
@@ -569,6 +569,7 @@
      "syncAnimatedImageToWordAudio": true, // For animated AVIF images, prepend a frozen first frame matching the existing word-audio duration so motion starts with sentence audio. Values: true | false
      "normalizeAudio": true, // Normalize generated sentence audio loudness during media extraction. Changes apply live. Values: true | false
      "mirrorMpvVolume": true, // Apply mpv's current software volume curve to generated sentence audio. Changes apply live. Values: true | false
+      "reviewTiming": false, // Review and preview subtitle media timing before SubMiner creates or enriches a mined card. Values: true | false
      "audioPadding": 0, // Seconds of padding appended to both ends of generated sentence audio and animated AVIF clips.
      "fallbackDuration": 3, // Fallback clip duration in seconds when subtitle timing data is unavailable.
      "maxMediaDuration": 30 // Maximum allowed media clip duration in seconds.
@@ -606,6 +607,11 @@
      "fieldGrouping": "disabled", // Kiku duplicate-card field grouping mode. Values: auto | manual | disabled
      "deleteDuplicateInAuto": true // When Kiku field grouping is "auto", delete the duplicate source card after grouping completes. Values: true | false
    }, // Is kiku setting.
+    "isSenren": {
+      "enabled": false, // Enable Senren-specific duplicate handling (scene-switching field grouping, including miscInfo grouping). Mutually exclusive with isKiku.enabled. Values: true | false
+      "fieldGrouping": "auto", // Senren duplicate-card field grouping mode (scene switching). Values: auto | manual | disabled
+      "deleteDuplicateInAuto": true // When Senren field grouping is "auto", delete the duplicate source card after grouping completes. Values: true | false
+    }, // Is senren setting.
    "lapisKiku": {
      "wordCardKind": "word-and-sentence" // Card-type flag SubMiner marks on Kiku/Lapis word cards. Only one flag is set at a time; the others are cleared. Requires isKiku.enabled or isLapis.enabled. Values: word-and-sentence | click | sentence | audio | none
    } // Lapis kiku setting.
@@ -136,6 +136,8 @@ SubMiner maps its data to your Anki note fields. Configure these under `ankiConn

 Field names are matched against your Anki note type case-insensitively (an exact match wins, then a lowercase comparison). If a configured field does not exist on the note type, SubMiner skips it without error.

+These mappings always control normal word-card enrichment, including Yomitan proxy/polling updates and manual clipboard updates. Enabling Lapis or Kiku does not replace the configured word-card sentence and audio fields with `Sentence` and `SentenceAudio`. The dedicated sentence-card and audio-card shortcuts still use those Lapis/Kiku field names.
+
 Two related options live alongside `fields`: `ankiConnect.deck` (target deck; empty falls back as described above) and `ankiConnect.tags` (tags added to mined cards, default `["SubMiner"]`; set `[]` to disable tagging). The `miscInfo` content is controlled by `ankiConnect.metadata.pattern` (default `[SubMiner] %f (%t)`; tokens: `%f` filename, `%F` filename with extension, `%t` timestamp, `%T` timestamp with milliseconds, `<br>` newline).

 ### Minimal Config
@@ -156,6 +158,8 @@ If you only want sentence and audio on your cards:

 SubMiner uses FFmpeg to generate audio and image media from the video. FFmpeg must be installed and on `PATH`.

+For remote streams such as Jellyfin playback, SubMiner downloads the clip's time window once into a temporary Matroska file (a stream copy, no re-encoding) and reads the timing review waveform, audio preview, audio, and image from that file instead of fetching the stream again for each step. The window covers the clip plus padding, plus the visible timeline in timing review, and grows when you reveal more of the timeline. It is deleted when a different window replaces it, after ten minutes without use, or when SubMiner exits. If the download fails, media generation reads the remote stream directly as before.
+
 ### Audio

 Audio is extracted from the video file using the subtitle's start and end timestamps. Padding is opt-in; keep it at `0` when you want sentence audio to start exactly at the mined sentence.
@@ -166,6 +170,7 @@ Audio is extracted from the video file using the subtitle's start and end timest
    "generateAudio": true,
    "normalizeAudio": true,      // normalize generated clip loudness
    "mirrorMpvVolume": true,     // apply the current mpv volume level
+    "reviewTiming": false,       // review and adjust timing before media generation
    "audioPadding": 0,           // optional seconds before and after subtitle timing
    "maxMediaDuration": 30       // cap total duration in seconds
  }
@@ -178,6 +183,12 @@ Output format: MP3 at 44100 Hz. If the video has multiple audio streams, SubMine

 The audio is uploaded to Anki's media folder and inserted as `[sound:audio_<timestamp>.mp3]`.

+Set `media.reviewTiming` to `true` to pause playback and review each word, sentence, or audio card before its media is generated. The review opens with the subtitle range plus configured audio padding. Subtitles usually linger past the dialogue, so once the waveform loads an untouched clip end moves back to just after the line's last speech (plus the configured padding); the Line end rail keeps marking the subtitle timing, Reset restores it, and a line whose speech runs through its end is left alone. Drag either edge of the clip to trim it, drag the middle to slide it without changing its length, or press anywhere else on the waveform to snap the nearer edge there. A focused edge also moves with the arrow keys, by 100 ms alone or 500 ms with Shift, and the 100 ms buttons do the same. Space previews the selection with a playhead that sweeps the clip; the preview ends when the hidden player has actually played the last sample, so output latency such as Bluetooth headphones does not cut the clip short. Enter confirms, and Escape cancels. The Earlier and Later buttons reveal another two seconds of available timeline without moving the selected clip. A speech-weighted waveform shows the mined subtitle as a tinted band with labeled line-start and line-end rails, making adjacent dialogue easier to distinguish. SubMiner uses a center channel when one carries dialogue, then falls back to a mono mix, keeps only the 250 to 3500 Hz speech band, and draws each slice's loudness relative to the clip's own noise floor, so steady background music or ambience reads as a flat line while dialogue stands out. Waveform analysis failure leaves the timing controls available. The confirmed range is exact: SubMiner does not apply audio padding a second time. Static screenshots use its midpoint, and animated AVIF clips use the full confirmed range.
+
+The review can also pull adjacent subtitle lines onto the card. Press `P` or `N` (or use the Prev and Next steppers above the sentence preview) to add the previous or next line, as many times as lines are available; Shift+`P` and Shift+`N` remove them again. The sentence preview lists every included line with the mined line highlighted, so the card's sentence field is always visible before you confirm, and the clip start or end, along with the line-start and line-end rails on the waveform, follows the outermost added line, keeping the review's audio padding. Confirming writes the combined lines to the sentence field; the Reset button drops the added lines along with any timing changes. Adjacent lines come from the parsed subtitle track when one is loaded; otherwise only lines that already played are offered, and a clip capped by `media.maxMediaDuration` keeps the full combined sentence even when the audio cannot cover every added line.
+
+Canceling the review lets you keep editing, finish with the original timing, keep or create the card without audio or an image, or discard the card. Discard deletes an existing Yomitan or audio card and skips creation for a direct sentence card. Clipboard updates and stats-dashboard mining do not open timing review. Audio preview failure does not block confirmation or card creation. The option is disabled by default and hot-reloads. You can also toggle **Review Media Timing** for the current session from the runtime options palette (`Ctrl/Cmd+Shift+O`).
+
 ### Screenshots (Static)

 A single frame is captured at the current playback position.
@@ -233,7 +244,7 @@ Animated AVIF requires an AV1 encoder (`libaom-av1`, `libsvtav1`, or `librav1e`)

 When media is available, mined-card overlay and system notifications include the same current-frame thumbnail.

-`overwriteAudio` applies to automatic card updates and duplicate-card enrichment. Manual clipboard subtitle updates (`Ctrl/Cmd+C`, then `Ctrl/Cmd+V`) always replace generated sentence audio, while leaving the word audio field unchanged.
+`overwriteAudio` applies to automatic card updates and duplicate-card enrichment. Manual clipboard subtitle updates (`Ctrl/Cmd+C`, then `Ctrl/Cmd+V`) always replace generated sentence audio in `ankiConnect.fields.audio`, even when `overwriteAudio` is disabled.

 ## AI Translation

@@ -287,6 +298,8 @@ Sentence card creation and audio card marking require a non-empty `ankiConnect.i

 Trigger with the mine sentence shortcut (`Ctrl/Cmd+S` by default). The card is created directly via AnkiConnect with the sentence, audio, and image filled in.

+The dedicated sentence-card and audio-card shortcuts use the Lapis/Kiku-compatible `Sentence` and `SentenceAudio` fields. This does not affect the configured fields used to enrich normal word cards.
+
 To mine multiple subtitle lines as one sentence card, use `Ctrl/Cmd+Shift+S` followed by a digit (1–9) to select how many recent lines to combine.

 ## Word Card Type (Kiku/Lapis)
@@ -304,9 +317,9 @@ Word cards get a card-type flag when SubMiner fills their sentence, whether that

 `click` marks `IsClickCard`, `sentence` marks `IsSentenceCard`, `audio` marks `IsAudioCard`, and `none` leaves the flags untouched for templates that manage them elsewhere. Whichever flag is chosen, the other card-type flags are cleared so the note never claims two card types. The setting is only read when `isKiku` or `isLapis` is enabled, and cards mined with Mine Sentence or Mine Audio keep their own flag.

-## Field Grouping (Kiku)
+## Field Grouping (Kiku/Senren)

-When you mine the same word multiple times, SubMiner can merge the cards instead of creating duplicates. This is designed for note types like [Kiku](https://github.com/youyoumu/kiku) that support grouped sentence/audio/image fields.
+When you mine the same word multiple times, SubMiner can merge the cards instead of creating duplicates. This is designed for note types that support grouped fields: [Kiku](https://github.com/youyoumu/kiku) and [Senren](https://github.com/BrenoAqua/Senren) (which calls the feature scene switching).

 ```jsonc
 "ankiConnect": {
@@ -318,6 +331,18 @@ When you mine the same word multiple times, SubMiner can merge the cards instead
 }
 ```

+For Senren note types, enable `isSenren` instead. Kiku and Senren write incompatible markup into the same fields, so only one can be enabled at a time; if both are enabled, Kiku wins and a config warning is emitted.
+
+```jsonc
+"ankiConnect": {
+  "isSenren": {
+    "enabled": true,
+    "fieldGrouping": "auto",           // "auto" (default), "manual", or "disabled"
+    "deleteDuplicateInAuto": true      // delete new card after auto-merge
+  }
+}
+```
+
 ### Modes

 **Disabled** (`"disabled"`): No duplicate detection. Each card is independent.
@@ -328,14 +353,17 @@ When you mine the same word multiple times, SubMiner can merge the cards instead

 ### What Gets Merged

-| Field    | Merge behavior                                |
-| -------- | --------------------------------------------- |
-| Sentence | Both cards' sentences kept as grouped entries |
-| Audio    | Both cards' `[sound:...]` entries kept        |
-| Image    | Both cards' images kept                       |
+| Field    | Merge behavior                                  |
+| -------- | ----------------------------------------------- |
+| Sentence | Both cards' sentences kept as grouped entries   |
+| Audio    | Both cards' `[sound:...]` entries kept          |
+| Image    | Both cards' images kept                         |
+| MiscInfo | Both cards' source info kept as grouped entries |

 Identical values from both cards are kept as separate grouped entries; the merge does not deduplicate.

+The merge markup depends on the note type. Kiku entries are wrapped in `<span data-group-id="...">` spans ordered newest first. Senren entries follow the [scene switching](https://github.com/BrenoAqua/Senren/blob/main/docs/scene_switching.md) format: sentence, sentenceFurigana, and miscInfo entries use `group` spans when ordinal order is sufficient and numbered `groupN` spans when they need an absolute scene target. Audio and pictures are appended positionally, and the number of sentenceAudio entries drives Senren's scene count. Ungrouped legacy content is wrapped into a group span on first merge, and source `groupN` spans are rebased after the kept note's existing audio scenes.
+
 ### Keyboard Shortcuts in the Modal

 | Key         | Action                             |
@@ -1,5 +1,92 @@
 # Changelog

+## v0.19.6 (2026-09-04)
+
+**Added**
+
+- **Card Timing Review**:
+  - Optional pre-generation timing review for word, sentence, and audio cards, with a speech-weighted waveform that flattens background noise so dialogue edges stand out clearly.
+  - The clip end automatically snaps back to where the line's dialogue actually ends once the waveform loads, with drag and keyboard adjustments available.
+  - Audio preview includes a sweeping playhead that plays the clip to its true end, even on high-latency outputs like Bluetooth headphones.
+  - Previous and next subtitle lines can be pulled onto the card with `P`/`N` (or the Prev/Next steppers) and removed with Shift; the sentence preview and waveform markers update automatically.
+  - Cancelling lets you keep a card without media, and the review can be toggled on or off for the session.
+- **Senren Field Grouping**:
+  - Enable `ankiConnect.isSenren` to merge duplicate mined cards using Senren's scene-switching markup, grouping sentence, furigana, audio, picture, and misc-info fields.
+  - Supports the same auto/manual/disabled modes as Kiku, including the manual merge modal; only one of Senren or Kiku can be enabled at a time.
+
+**Changed**
+
+- **Remote Stream Mining Performance**: Mining a card from a remote stream (Jellyfin and other HTTP sources) now downloads the clip window once and reuses it for the timing review waveform, audio preview, audio extraction, and screenshot, instead of re-fetching the stream at each step; the temporary file is cleaned up after ten minutes of inactivity or on exit.
+- **TsukiHime Release Filtering**: The TsukiHime modal's Japanese and secondary-language tabs now filter the release list by the subtitle languages each release actually carries, and report when no release has subtitles for the active tab.
+
+**Fixed**
+
+- **Subtitle & Mining Accuracy**:
+  - Broadcast-style captions that split one sentence across two on-screen rows (e.g. Crunchyroll Japanese subs) now merge into a single line for the sidebar and mined cards, while separate speakers, sound effects, and labeled turns still stay on their own lines.
+  - Mining from the overlay no longer pulls in a lingering row from the previous caption; the mined sentence and clip timing now match what's actually on screen.
+  - Multi-line copy and mining now select lines backward in timeline order after seeking, instead of in playback encounter order.
+  - Copying a subtitle, mining a sentence, or recording immersion stats no longer includes the separate furigana line that broadcast ASS captions place above a word.
+- **Card Update Notifications**: Dismissed lingering overlay card-update progress when notification settings switch to OSD before an update finishes.
+- **Overlay Stability on Hyprland**: Opening a modal window (timing review, Jimaku, session help, and others) while mpv is fullscreen no longer causes the overlay to flicker while the modal loads; the overlay now stays on screen untouched until the modal is ready.
+- **Jellyfin Subtitle Sync**: Jellyfin subtitle files now load with zero mpv delay instead of inferring and saving an offset from Japanese and English cue timelines.
+- **Secondary Subtitle Visibility**: Native mpv secondary subtitles stay hidden when switching secondary subtitle tracks during playback.
+
+## v0.19.5 (2026-08-30)
+
+**Fixed**
+
+- **Anki Card Update Progress**: The card-update spinner now stays visible until audio and image updates finish, instead of disappearing early.
+- **Anki Word-Card Fields**: Word-card enrichment now writes sentence text and audio to the fields configured in AnkiConnect, while the dedicated sentence-card and audio-card actions keep their existing compatible field names.
+- **Overlapping Subtitles**:
+  - Subtitle lines that start while another line is still on screen now appear alongside it, instead of staying hidden until a track switch or seek.
+  - Subtitles shown at the same time now stack by their authored screen position, with top signs and song lines above bottom dialogue.
+  - Half-size ASS furigana is no longer shown as if it were a dialogue line.
+- **YouTube Auto Captions**:
+  - Auto-generated captions now follow their intended timing and two-row roll-up layout.
+  - Long speech is paged instead of covering the video with a wall of text.
+  - Explicitly timed sound cues like `[音楽]` no longer cover later dialogue.
+
+## v0.19.4 (2026-08-25)
+
+**Added**
+- **Library Merge & Move**: Duplicate library cards for the same show can now be combined. Select cards in the library grid and use "Merge Selected" to pick which entry to keep and move every episode onto it, preserving sessions, mined cards, and watch time. Episodes can also be reassigned individually via the "→" button, useful when a file lands under a stray title; manual assignments survive later filename parsing, Jellyfin refreshes, and season repair. Exact AniList title matches with compatible seasons now merge automatically, while fuzzy matches surface as dismissible "Possible duplicate" reviews instead of merging silently.
+- **Duplicate Line Cleanup Tool**: The Vocabulary tab's new "Duplicates" button scans a chosen time window (7 days through all time) for old karaoke/typeset duplicate-line bursts, shows what it found, and collapses each run to one line once confirmed; `subminer stats cleanup --duplicate-lines` does the same from the terminal, with `--dry-run` and `--lookback-days <n>` options. Watch time and lines-seen totals are left unchanged.
+
+**Changed**
+- **Prerelease Release Notes**: Prerelease notes now open with a "Changes since" section listing only what changed versus the previous beta/RC of the same version, above the cumulative highlights, and CI rejects prerelease tags whose committed notes were generated for a different beta/RC.
+
+**Fixed**
+- **Subtitle & Karaoke Duplication**:
+  - Karaoke and animated signs are reconstructed once from their authored text and shown only while actually sung, with original word spacing preserved, instead of flooding the overlay, subtitle sidebar, immersion history, mining, or stats with glyph fragments, per-frame color phases, and repeated animation events.
+  - Decorative layers (highlight sweeps, glow/shadow copies, symbol-font decoration, particle swarms, hidden or zero-scaled text) stay out of published text, while ordinary repeated dialogue, positioned signs, wrapped lyric rows, and multi-row CC-style blocks still display correctly.
+  - Embedded subtitle tracks on network-mounted (SMB/NFS) media are extracted and parsed again instead of falling back to live-text-only, restoring karaoke reconstruction, sidebar cues, and mining for releases that only ship subtitles inside the container.
+  - Secondary subtitles go through the same deduplication pipeline as primary subtitles and no longer clip display after about four lines.
+  - Event-heavy karaoke files that previously stalled subtitle loading for several seconds now parse in well under a second.
+- **Character Dictionary Reliability**:
+  - Generation, merged rebuilds, and imports no longer freeze the app on large dictionaries; snapshot I/O, archive building, and image/name lookup caches moved off the UI's critical path.
+  - Dictionaries are reused instead of regenerated when MeCab finds no name splits.
+  - Cached portraits restore correctly after the portrait index finishes loading post-tokenization.
+  - Desktop progress notifications on Linux AppImage installs update in place instead of flickering, fixing a bug where the AppImage's bundled libraries broke the system notification helper.
+- **Overlay Startup & Modals**:
+  - The macOS window-tracking helper targets macOS 12.0+ instead of requiring the build machine's exact macOS version, fixing crashes on older systems like Ventura that left the overlay stuck on "Overlay loading".
+  - mpv IPC connection attempts time out and retry, showing an actionable error if content still isn't ready after 30 seconds.
+  - Dedicated overlay modals are prewarmed on macOS and Windows so shortcuts open them promptly.
+  - On macOS, reused modals and the stats window open above fullscreen mpv on its current Space instead of jumping to another desktop.
+- **Wayland File Drop**: Fixed native Wayland drag-and-drop from file managers such as Thunar, so subtitle and video files dropped on the visible overlay are resolved and forwarded to mpv.
+- **Windows Mouse Lag**: Fixed system-wide mouse lag on Windows while SubMiner is running, caused by the overlay's global mouse hook for click-through forwarding and by the mpv window tracker blocking the app on repeated PowerShell lookups.
+- **Sentence Mining Audio & Clips**: Sentence-audio generation no longer times out on slow network-mounted media with many subtitle/font streams (bounded FFmpeg probing, two-minute extraction budget, clearer error reporting), and mined audio/animated AVIF clips now capture the subtitle line that was actually mined by snapshotting the clip range at lookup time instead of reading live mpv state later.
+- **Stats Performance & Reliability**: Immersion stats storage now sets its SQLite busy timeout before WAL setup, avoiding transient lock errors under concurrent writes. Deletes in the stats dashboard no longer freeze the UI, run proportional to what's deleted instead of rebuilding full lifetime summaries, retry safely if the delete worker crashes, and no longer rescan the whole library when deleting very common words; a new index also makes large session deletes drop from minutes to milliseconds. Library merges, video moves, and AniList reassignments got the same lifetime-summary fix.
+- **Vocabulary Stats Accuracy**: Vocabulary totals and charts now count all tracked vocabulary instead of only the first page, new-word history uses corrected daily rollups (fixing legacy timestamp and time-zone issues), summary cards refresh automatically after edits to the exclusion list, and rapid exclusion edits no longer race each other.
+- **Rofi MKV Thumbnails**: Fixed missing MKV thumbnails in the Linux rofi picker when system thumbnailer registrations only advertise legacy Matroska MIME aliases.
+
+<details>
+<summary>Internal changes</summary>
+
+**Internal**
+- Docs Site Indexing: Excluded the `/main/` and `/v/<version>/` docs trees from search indexing (self-referential canonical, `noindex,follow`, matching `X-Robots-Tag`) so crawlers focus on current docs instead of ~30 archived copies of every page, and restored `<lastmod>` dates in the docs sitemap that were silently dropped by production builds.
+
+</details>
+
 ## v0.19.3 (2026-08-13)

 **Added**
@@ -148,9 +148,9 @@ The configuration file includes several main sections:

 - [**Shared AI Provider**](#shared-ai-provider) - Canonical OpenAI-compatible provider config shared by Anki and YouTube subtitle fixing
 - [**AnkiConnect**](#ankiconnect) - Automatic Anki card creation with media
- [**Kiku/Lapis Integration**](#kiku-lapis-integration) - Sentence cards and duplicate handling for Kiku/Lapis note types
+- [**Kiku/Lapis Integration**](#kiku-lapis-integration) - Sentence cards and duplicate handling for Kiku/Lapis/Senren note types
 - [**N+1 Word Highlighting**](#n-1-word-highlighting) - Known-word cache and single-target highlighting
- [**Field Grouping Modes**](#field-grouping-modes) - Kiku/Lapis duplicate card merging
+- [**Field Grouping Modes**](#field-grouping-modes) - Kiku/Senren duplicate card merging

 **External Integrations**

@@ -873,9 +873,10 @@ When config hot-reload updates shortcut/keybinding/style values, close and reope

 Use the runtime options palette to toggle settings live while SubMiner is running. These changes are session-only and reset on restart.

-Current runtime options cover automatic card updates, known-word highlighting,
-known-word maturity coloring, N+1 annotation, JLPT underlines, frequency
-highlighting, known-word match mode, and Kiku field grouping mode.
+Current runtime options cover automatic card updates, media timing review,
+known-word highlighting, known-word maturity coloring, N+1 annotation, JLPT
+underlines, frequency highlighting, known-word match mode, and Kiku field
+grouping mode.

 Annotation toggles only apply to new subtitle lines after the toggle. The currently displayed line is not re-tokenized in place.

@@ -967,6 +968,7 @@ Enable automatic Anki card creation and updates with media generation:
      "animatedCrf": 35,
      "normalizeAudio": true,
      "mirrorMpvVolume": true,
+      "reviewTiming": false,
      "audioPadding": 0,
      "fallbackDuration": 3,
      "maxMediaDuration": 30
@@ -1019,6 +1021,7 @@ This example is intentionally compact. The option table below documents availabl
 | `media.generateAudio`                             | `true`, `false`                             | Generate audio clips from video (default: `true`)                                                                                                                                                                               |
 | `media.normalizeAudio`                            | `true`, `false`                             | Normalize generated sentence-audio loudness during media extraction (default: `true`). Set to `false` to keep raw source loudness. Changes apply live.                                                                          |
 | `media.mirrorMpvVolume`                           | `true`, `false`                             | Apply mpv's cubic software-volume curve to each generated sentence-audio clip (default: `true`). This ignores mpv's separate mute state, falls back to unity scaling if volume cannot be read, and applies changes live.        |
+| `media.reviewTiming`                              | `true`, `false`                             | Pause playback and review word, sentence, and audio card timing before media generation (default: `false`). Clipboard updates and stats-dashboard mining do not open the review.                                                |
 | `media.generateImage`                             | `true`, `false`                             | Generate image/animation screenshots (default: `true`)                                                                                                                                                                          |
 | `media.imageType`                                 | `"static"`, `"avif"`                        | Image type: static screenshot or animated AVIF (default: `"static"`)                                                                                                                                                            |
 | `media.imageFormat`                               | `"jpg"`, `"png"`, `"webp"`                  | Image format (default: `"jpg"`)                                                                                                                                                                                                 |
@@ -1051,6 +1054,7 @@ This example is intentionally compact. The option table below documents availabl
 | `metadata.pattern`                                | string                                      | Format pattern for metadata: `%f`=filename, `%F`=filename+ext, `%t`=time, `%T`=time with milliseconds, `<br>`=newline                                                                                                           |
 | `isLapis`                                         | object                                      | Lapis/shared sentence-card config: `{ enabled, sentenceCardModel }`. Sentence/audio field names are fixed to `Sentence` and `SentenceAudio`.                                                                                    |
 | `isKiku`                                          | object                                      | Kiku-only config: `{ enabled, fieldGrouping, deleteDuplicateInAuto }` (shared sentence/audio/model settings are inherited from `isLapis`)                                                                                       |
+| `isSenren`                                        | object                                      | Senren-only config: `{ enabled, fieldGrouping, deleteDuplicateInAuto }`. Merges duplicates using Senren's scene-switching markup. Mutually exclusive with `isKiku.enabled`.                                                      |

 `ankiConnect.ai` only controls feature-local enablement plus optional `model` / `systemPrompt` overrides.
 API key resolution, base URL, and timeout live under the shared top-level [`ai`](#shared-ai-provider) config.
@@ -1080,6 +1084,7 @@ SubMiner is intentionally built for [Kiku](https://kiku.youyoumu.my.id/) and [La
 - Enable `isKiku` to turn on duplicate merge behavior for mined Word/Expression hits.
 - When both are enabled, Kiku behavior is applied for grouping while sentence-card model settings are still read from `isLapis`.
 - `isKiku.fieldGrouping` supports `disabled`, `auto`, and `manual` merge modes; see [Field Grouping Modes](#field-grouping-modes).
+- For [Senren](https://github.com/BrenoAqua/Senren) note types, enable `isSenren` instead of `isKiku`. Duplicate merges then use Senren's scene-switching markup (including grouped `miscInfo` entries), and `isSenren.fieldGrouping` supports the same three modes (default: `auto`). Kiku and Senren are mutually exclusive; if both are enabled, Kiku wins and Senren is turned off with a config warning.
 - `lapisKiku.wordCardKind` picks the card-type flag set on word cards; see [Word Card Type](#word-card-type). It is read only while `isLapis` or `isKiku` is enabled.

 ### Word Card Type
@@ -6,7 +6,7 @@ For internal architecture/workflow guidance, use `docs/README.md` at the repo ro

 - [Bun](https://bun.sh)
 - A system `lua` interpreter for `bun run test:launcher` / `bun run test:plugin:src`
- macOS builds compile a Swift helper via `scripts/build-macos-helper.sh` (skip with `SUBMINER_SKIP_MACOS_HELPER_BUILD=1`)
+- macOS builds compile a Swift helper via `scripts/prepare-build-assets.mjs` (skip with `SUBMINER_SKIP_MACOS_HELPER_BUILD=1`)

 ## Setup

@@ -54,7 +54,7 @@ From then on, pause / resume / seek / stop and audio or subtitle track changes y
 - **Resume works.** If Jellyfin has a saved position for the item, SubMiner seeks there on load.
 - **Direct play first.** When the source allows it and the container is in your direct-play allowlist, SubMiner streams the original file; otherwise it requests a transcoded stream from Jellyfin.
 - **Japanese subtitles are auto-selected,** preferring Jellyfin's default and embedded tracks over external sidecar files when several match.
- **Subtitle timing is corrected when possible.** SubMiner removes Jellyfin's server-selected subtitle stream from the mpv load URL, suppresses the mpv plugin's one-shot subtitle auto-selection and overlay auto-start for managed Jellyfin loads, stages downloaded subtitle tracks without letting mpv auto-switch between tracks, then selects the Japanese track once after applying any saved or inferred timing delay. When Jellyfin provides both Japanese and English subtitle files, SubMiner compares their cue timelines and applies a global delay if one track is clearly offset. Manual delay shifts you make with SubMiner's adjacent-cue controls are saved per item and subtitle track, then restored the next time you select that track.
+- **Downloaded subtitles keep their original timing.** SubMiner removes Jellyfin's server-selected subtitle stream from the mpv load URL, suppresses the mpv plugin's one-shot subtitle auto-selection and overlay auto-start for managed Jellyfin loads, stages the subtitle files exposed by Jellyfin without letting mpv auto-switch between tracks, resets mpv's subtitle delay to zero, then selects the Japanese track. SubMiner does not compare Japanese and English cue timelines or save an inferred delay.

 ## Settings

@@ -41,7 +41,7 @@ If you prefer a hands-on approach (animecards-style), you can copy the current s
   - For multiple lines: press `Ctrl/Cmd+Shift+C`, then a digit `1`–`9` to select how many recent subtitle lines to combine. The combined text is copied to the clipboard.
 3. Press `Ctrl/Cmd+V` to update the last-added card with the clipboard contents plus audio, image, and translation - the same fields auto-update would fill.

-Manual clipboard updates always replace generated sentence audio, even when `ankiConnect.behavior.overwriteAudio` is disabled. The word audio field is left unchanged because the word itself does not change in this flow.
+Manual clipboard updates always replace generated sentence audio in `ankiConnect.fields.audio`, even when `ankiConnect.behavior.overwriteAudio` is disabled. Normal word-card updates use the configured sentence and audio fields even when Lapis or Kiku support is enabled.

 This is useful when auto-update is disabled or when you want explicit control over which subtitle line gets attached to the card.

@@ -72,17 +72,17 @@ After adding a word via Yomitan, press the audio card shortcut (`Ctrl/Cmd+Shift+
 Audio card marking uses the same `ankiConnect.isLapis.sentenceCardModel` note type as sentence cards. See [Anki Integration - Sentence Cards](/anki-integration#sentence-cards-lapis) for setup.
 :::

-### Field Grouping (Kiku)
+### Field Grouping (Kiku/Senren)

-If you mine the same word from different sentences, SubMiner can merge the cards instead of creating duplicates. This feature is designed for use with [Kiku](https://github.com/youyoumu/kiku) and similar note types that support grouped fields.
+If you mine the same word from different sentences, SubMiner can merge the cards instead of creating duplicates. This feature is designed for use with [Kiku](https://github.com/youyoumu/kiku) and [Senren](https://github.com/BrenoAqua/Senren) note types that support grouped fields (Senren calls it scene switching).

 1. You add a word via Yomitan.
 2. SubMiner detects the new card and checks if a card with the same expression already exists.
-3. If a duplicate is found (this requires `ankiConnect.isKiku.fieldGrouping` to be set to `"auto"` or `"manual"`; it defaults to `"disabled"`):
-   - **Auto mode** (`ankiConnect.isKiku.fieldGrouping: "auto"`): Merges automatically. Both sentences, audio clips, and images are combined into the existing card. The duplicate is optionally deleted.
-   - **Manual mode** (`ankiConnect.isKiku.fieldGrouping: "manual"`): A modal appears showing both cards side by side. You choose which card to keep and preview the merged result before confirming.
+3. If a duplicate is found (this requires Kiku or Senren to be enabled with a field grouping mode of `"auto"` or `"manual"`):
+   - **Auto mode**: Merges automatically. Both sentences, audio clips, images, and source info are combined into the existing card. The duplicate is optionally deleted.
+   - **Manual mode**: A modal appears showing both cards side by side. You choose which card to keep and preview the merged result before confirming.

-See [Anki Integration - Field Grouping](/anki-integration#field-grouping-kiku) for configuration options, merge behavior, and modal keyboard shortcuts.
+See [Anki Integration - Field Grouping](/anki-integration#field-grouping-kiku-senren) for configuration options, merge behavior, and modal keyboard shortcuts.

 ## Overlay Model

@@ -108,8 +108,12 @@ The secondary bar is a compact top-strip region in the same overlay window. It s
 - Quick comprehension checks without leaving the mining flow.
 - Auto-populating the translation field on mined cards - when a card is created, SubMiner uses the secondary subtitle text as the translation field value (unless AI translation is configured to override it).

+For local media, SubMiner can parse supported embedded secondary tracks into timed cues. For remote URLs and files on network mounts, it uses mpv's live secondary subtitle text instead of scanning the media with ffmpeg.
+
 It is controlled by `secondarySub` configuration and shares its lifecycle with the main overlay window. Cycle which track feeds it with `Shift+J`.

+SubMiner collapses duplicate ASS layers in parsed secondary tracks. Exact repeated lines collapse at any length, while distinct simultaneous short lines remain separate. Long dialogue and positioned-sign copies also collapse when they differ only in whitespace or terminal punctuation. Dense multi-row sign layouts, such as translated timetables, are excluded instead of being concatenated into the secondary bar.
+
 ### Display Modes

 Both the primary and secondary subtitle bars share the same three visibility modes, and each can be changed independently at runtime:
@@ -523,7 +523,7 @@
  // ==========================================
  // AnkiConnect Integration
  // Automatic Anki updates and media generation options.
-  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.
+  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume/reviewTiming, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, isSenren.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.
  // Shared AI provider transport settings are read from top-level ai and typically require restart.
  // Most other AnkiConnect settings still require restart.
  // ==========================================
@@ -569,6 +569,7 @@
      "syncAnimatedImageToWordAudio": true, // For animated AVIF images, prepend a frozen first frame matching the existing word-audio duration so motion starts with sentence audio. Values: true | false
      "normalizeAudio": true, // Normalize generated sentence audio loudness during media extraction. Changes apply live. Values: true | false
      "mirrorMpvVolume": true, // Apply mpv's current software volume curve to generated sentence audio. Changes apply live. Values: true | false
+      "reviewTiming": false, // Review and preview subtitle media timing before SubMiner creates or enriches a mined card. Values: true | false
      "audioPadding": 0, // Seconds of padding appended to both ends of generated sentence audio and animated AVIF clips.
      "fallbackDuration": 3, // Fallback clip duration in seconds when subtitle timing data is unavailable.
      "maxMediaDuration": 30 // Maximum allowed media clip duration in seconds.
@@ -606,6 +607,11 @@
      "fieldGrouping": "disabled", // Kiku duplicate-card field grouping mode. Values: auto | manual | disabled
      "deleteDuplicateInAuto": true // When Kiku field grouping is "auto", delete the duplicate source card after grouping completes. Values: true | false
    }, // Is kiku setting.
+    "isSenren": {
+      "enabled": false, // Enable Senren-specific duplicate handling (scene-switching field grouping, including miscInfo grouping). Mutually exclusive with isKiku.enabled. Values: true | false
+      "fieldGrouping": "auto", // Senren duplicate-card field grouping mode (scene switching). Values: auto | manual | disabled
+      "deleteDuplicateInAuto": true // When Senren field grouping is "auto", delete the duplicate source card after grouping completes. Values: true | false
+    }, // Is senren setting.
    "lapisKiku": {
      "wordCardKind": "word-and-sentence" // Card-type flag SubMiner marks on Kiku/Lapis word cards. Only one flag is set at a time; the others are cleared. Requires isKiku.enabled or isLapis.enabled. Values: word-and-sentence | click | sentence | audio | none
    } // Lapis kiku setting.
@@ -35,7 +35,7 @@ These work when the overlay window has focus.
 | `Ctrl/Cmd+G`       | Trigger field grouping (Kiku merge check)       | `shortcuts.triggerFieldGrouping`        |
 | `Ctrl/Cmd+Shift+A` | Mark last card as audio card                    | `shortcuts.markAudioCard`               |

-The multi-line shortcuts open a digit selector with a 3-second timeout (`shortcuts.multiCopyTimeoutMs`). Press `1`–`9` to select how many recent subtitle lines to combine. When the shortcut starts from mpv, SubMiner focuses the visible overlay for that selector instead of reserving the number keys in the mpv plugin.
+The multi-line shortcuts open a digit selector with a 3-second timeout (`shortcuts.multiCopyTimeoutMs`). Press `1`–`9` to select the total number of subtitle lines to combine, ending at the current line and moving backward through the subtitle timeline. The current line counts toward the selected total. When the shortcut starts from mpv, SubMiner focuses the visible overlay for that selector instead of reserving the number keys in the mpv plugin.

 ## Overlay Controls

@@ -9,7 +9,7 @@ The sidebar is enabled by default. Set `subtitleSidebar.enabled` to `false` if y
 When SubMiner parses the active subtitle source into a cue list, the sidebar becomes available. Toggle it with the `\` key (configurable via `subtitleSidebar.toggleKey`). While open:

 - The active cue is highlighted and kept in view as playback advances (when `autoScroll` is `true`).
- Clicking any cue seeks mpv to that timestamp.
+- Clicking any cue seeks mpv into that line. For overlapping ASS karaoke, SubMiner moves past the previous line's exit animation when the selected cue has enough time remaining.
 - The sidebar stays synchronized with the overlay - media transitions and subtitle source changes update both simultaneously.

 For typeset ASS karaoke and animated signs, SubMiner collapses generated animation frames and repeated full-line color phases before they reach the sidebar. It recovers a clean complete line from a matching timed authoring comment or from full-line events surrounding generated fragments. Ordinary ASS comments, editor notes, alternate lines, repeated dialogue, and separately positioned signs remain distinct.
@@ -14,7 +14,7 @@ Unlike Jimaku, TsukiHime needs no account or API key. The only requirement is th

 ## How It Works

-The integration runs through an in-overlay modal opened with `Ctrl+Shift+T` by default. The modal has two tabs that filter the subtitle tracks of the selected release by role: the first follows `secondarySub.secondarySubLanguages` (English when unset), and the second is always **Japanese**, the currently supported primary subtitle language. Tracks with no language tag stay visible on the secondary tab.
+The integration runs through an in-overlay modal opened with `Ctrl+Shift+T` by default. The modal has two tabs that filter both the release list and the subtitle tracks of the selected release by role: the first follows `secondarySub.secondarySubLanguages` (English when unset), and the second is always **Japanese**, the currently supported primary subtitle language. Each tab lists only the releases whose reported subtitle languages include the tab's language, so the Japanese tab hides the many releases that ship English subtitles only. Releases and tracks with no language tag stay visible on the secondary tab. If nothing on the active tab qualifies, the status line says so and points at the other tab.

 When you open the modal, SubMiner parses the current video filename to extract a title and episode number (same parser as Jimaku - `S01E03`, `1x03`, `E03`, and dash-separated numbers all work). If the filename yields a high-confidence match, SubMiner auto-searches immediately.

@@ -76,6 +76,7 @@ The previous `--open-animetosho` flag and `__animetosho-open` keybinding command
 ## Troubleshooting

 - **"xz binary not found"** - install `xz`/`xz-utils` with your package manager.
+- **"No releases with Japanese subtitles"** - none of the search results carry a Japanese track. Most releases only ship English subtitles; try another search, or use the [Jimaku integration](/jimaku-integration) for Japanese subtitles.
 - **"Batch releases are not supported"** - TsukiHime only exposes extracted attachments for single-file torrents. Pick the single-episode release for your episode instead of a season batch.
 - **"No text subtitle tracks in this release"** - the release only carries image-based subtitles (PGS/VobSub) or none at all; try a different release (fansub and SubsPlease-style releases almost always carry ASS tracks).
 - **Timing is off** - the subtitle came from a different release than your video file. Use the subtitle sync modal (`Ctrl+Alt+S`) or pick the release matching your file exactly.
@@ -58,12 +58,15 @@
   `latest*.yml` and `*.blockmap` files under `release/`.
 5. Commit the prerelease prep (package.json version bump + the generated
   `release/prerelease-notes.md`). CI does not regenerate notes — it uses the
-   committed file — so review it before committing. If you add more
-   `changes/*.md` fragments for a later beta/RC, rerun
-   `bun run changelog:prerelease-notes --version <version>`; the generator uses
-   the existing prerelease notes as the baseline only when their hidden
-   `prerelease-base-version` marker matches the current base version, and asks
-   Claude to merge only the new fragment material. Do not run
+   committed file — so review it before committing. Rerun
+   `bun run changelog:prerelease-notes --version <version>` for every later
+   beta/RC, even if no fragments changed: the notes carry a hidden
+   `prerelease-version` marker and CI rejects the tag when the marker does not
+   match it (verify locally with
+   `bun run changelog:check-prerelease-notes --version <version>`). The
+   generator reuses the existing notes as the cumulative baseline when their
+   marker (or legacy `prerelease-base-version` marker) matches the current base
+   version, and asks Claude to merge only the new fragment material. Do not run
   `bun run changelog:build`.
 6. Tag the commit: `git tag v<version>`.
 7. Push commit + tag.
@@ -78,6 +81,8 @@ Notes:
 - Pass `--date` explicitly when you want the release stamped with the local cut date; otherwise the generator uses the current ISO date, which can roll over to the next UTC day late at night.
 - `changelog:check` now rejects tag/package version mismatches.
 - `changelog:prerelease-notes` also rejects tag/package version mismatches and writes `release/prerelease-notes.md` without mutating tracked changelog files. When that file already exists, the generator includes it in the Claude prompt so later beta/RC notes reuse the reviewed text instead of starting over.
+- From the second prerelease of a base version onward, the notes open with a `## Changes since <previous tag>` section above the cumulative `## Highlights`. The generator locates the newest preceding beta/RC tag for the same base version (semver order: all betas before all RCs), diffs `changes/*.md` between that tag and the working tree, and asks Claude to describe only the behavioral beta-to-beta differences — added fragments as new changes, modified fragments by their before/after difference (editorial-only edits are dropped), deleted fragments as removed/reverted changes. If no fragments changed (for example a packaging-only rebuild), the section states that explicitly without a Claude call. The delta section carries no separate contributor attribution; `## What's Changed` stays cumulative like `## Highlights`.
+- `changelog:check-prerelease-notes --version <version>` verifies the committed notes' `prerelease-version` marker matches the version being tagged; the prerelease workflow runs it and fails the release on stale notes.
 - `changelog:build` generates `CHANGELOG.md` + `release/release-notes.md` (both polished by `claude -p`) and removes the released `changes/*.md` fragments. The CHANGELOG keeps internal notes inside a `<details><summary>Internal changes</summary>` collapse; the release notes drop them entirely.
 - `release/release-notes.md` (and `release/prerelease-notes.md`) include GitHub-style attribution after `## Highlights`: a `## What's Changed` list crediting each released fragment as `by @<author> in #<pr>`, plus a `## New Contributors` section for first-time authors. Attribution is resolved per fragment via `git log` (the commit that added the fragment) + `gh api .../commits/<sha>/pulls`, with one `gh` search per author for the first-contribution check. It needs `gh` installed and authenticated; if `gh` is unavailable or a lookup fails, the generator warns and emits notes without the attribution sections rather than failing. The CHANGELOG itself stays attribution-free.
 - The release workflow no longer auto-runs `changelog:build`. If pending `changes/*.md` fragments are present on a tag-based run, CI exits with a clear `::error::` pointing at the local fix. Run `bun run changelog:build --version <version>` locally, commit the polished output, then tag.
@@ -87,7 +87,9 @@ interface SubtitleCue {

 ASS scripts can also redraw one complete lyric for two or more long color/highlight phases. Those flush-timed phases collapse separately from short animation frames when they share text, style, actor, and layer and carry direct animation evidence, such as temporal tags or changing non-spatial overrides. Spatial command changes do not prove a phase, so separately positioned signs remain distinct.

-**Canonical animation recovery.** Some ASS producers keep the readable lyric or sign as a timed `Comment:` and generate hundreds of `Dialogue:` frames containing repeated glyphs or changing clip regions. Others retain the complete line as brief `Dialogue:` events around the generated fragments. A complete event is promoted only when nearby dialogue from the same style and actor forms a proven animation cluster and reconstructs its entire text in source order. The generated frames are then replaced by one cue marked `source: 'canonical-ass'`. This source marker lets the live primary-subtitle path prefer the clean authored text and timing for display, sidebar history, immersion recording, and mining, while unmatched editor notes and alternative translations remain ignored.
+**Canonical animation recovery.** Some ASS producers keep the readable lyric or sign as a timed `Comment:` and generate hundreds of `Dialogue:` frames containing repeated glyphs or changing clip regions. Others retain the complete line as brief `Dialogue:` events around the generated fragments. A complete event is promoted only when nearby dialogue from the same style and actor forms a proven animation cluster and reconstructs its entire text in source order. The generated frames are then replaced by one cue marked `source: 'canonical-ass'`. This source marker lets the live primary-subtitle path prefer the clean authored text and timing for display, sidebar history, immersion recording, and mining, while unmatched editor notes and alternative translations remain ignored. Secondary selection advances to an entering canonical cue at its generated animation start when the preceding authored cue ends before the new authored span. Unrelated simultaneous cues that continue through the new span remain visible.
+
+**Font texture cleanup.** A clipped repeated-glyph run or frequent changes to secondary alpha marks a texture seed. Clipped runs do not need a font override because some signs build their masks from ordinary `l` glyphs. The parser removes short clipped pieces that share a no-font seed's style and timing, or pieces that share a font seed's style, timing, and font even when the actor changes. It also removes positioned text layers with at least `E0` global alpha when they overlap a seed in the same style. Opaque authored sign text stays publishable when the texture switches fonts or actors around it.

 #### Prefetch Service Lifecycle

@@ -3,7 +3,7 @@
 # Subtitle Overlay Priming

 Status: active
-Last verified: 2026-08-18
+Last verified: 2026-08-19
 Owner: Kyle Yasuda
 Read when: debugging subtitle state or blank Linux/X11 overlay windows when the visible overlay is shown or recreated

@@ -71,11 +71,24 @@ coming and prefetching would otherwise idle for the rest of the cue.

 - Primary live text first resolves recovered canonical ASS animations. Otherwise, when
  every live mpv line matches an active parsed cue, it uses the parsed cue text so exact
-  full-span style layers appear once instead of repeating for fill, border, blur, and
-  shadow events. Any unmatched live line keeps the complete live stack, preserving
-  dialogue or signs that overlap a lyric.
+  full-span style layers appear once instead of repeating for fill, border, blur, shadow,
+  or equivalent whitespace variants. Any unmatched live line keeps the complete live
+  stack, preserving dialogue or signs that overlap a lyric.
 - A tokenization cache miss emits the plain cue synchronously. Tokenization remains serialized so
  live work does not contend for Yomitan state.
+- The initial `time-pos`, explicit renderer seeks, and later seek-like jumps reprocess mpv's
+  current raw `sub-text` after the new playback time is stored. Explicit intent matters because
+  adjacent subtitle jumps can be shorter than the general seek-distance threshold. This corrects
+  ASS cleanup when mpv delivered the destination subtitle before the destination timestamp.
+- Renderer `sub-seek` commands use the active parsed cue list when available. Simultaneous cues
+  share one boundary, overlapping lyrics advance from the latest active boundary, and mpv's native
+  command remains the fallback when no parsed destination exists. This prevents generated karaoke
+  frames from consuming next/previous subtitle presses.
+- Subtitle sidebar selections seek past the preceding sanitized cue's overlapping exit span when
+  the selected cue has enough time remaining. This keeps direct row selection on the requested
+  karaoke line while clamping the seek inside that cue.
+- If startup paints raw text before embedded ASS parsing finishes, parsed cue arrival may replace
+  that provisional line. The one-prime-per-media guard still suppresses identical repeats.
 - If a newer cue arrives while an older line is still tokenizing, the newer plain cue or empty
  clear payload is emitted immediately. The older tokenization result is dropped before it can
  replace the current cue.
@@ -84,17 +97,63 @@ coming and prefetching would otherwise idle for the rest of the cue.

 ## Secondary Subtitle Flow

- `secondary-sub-text` remains the immediate fallback, so unreadable and remote subtitle sources
-  still appear without waiting for file resolution.
+- `secondary-sub-text` remains the immediate fallback, so unreadable subtitle sources, remote URLs,
+  and still-extracting embedded tracks appear without waiting for file resolution. Embedded-track
+  extraction runs for local and network-mounted files alike (demuxing reads the whole container,
+  about 10 seconds per GB on gigabit, under a generous timeout); only true remote URLs skip it,
+  having no on-disk container to demux.
+- The live fallback also suppresses per-glyph typesetting walls: when many simultaneous
+  one-glyph lines are present (generated karaoke lettering flattened into live text), those
+  lines and their short syllable companions are dropped while concurrent dialogue lines stay.
+  This keeps the overlay clean while extraction is still in flight and for sources that never
+  produce parsed cues.
+- Parsed secondary text and the live fallback remove exact repeated lines at any length. A
+  flattened-line identity also removes long dialogue/sign repetitions that differ only in
+  whitespace or terminal punctuation, while distinct simultaneous short lines remain separate.
 - `secondary-subtitle-track.ts` resolves `secondary-sid` against mpv's track list. External tracks
  are read directly; supported embedded text tracks are extracted through the same ffmpeg-backed
  source resolver used by primary subtitle prefetching.
 - The selected source is parsed with `parseSubtitleCues()`, including metadata-aware ASS duplicate
  and animation collapse. Playback `time-pos` selects the active parsed cue after applying
  `secondary-sub-delay`.
+- Fragment reconstruction marks tall multi-row positioned parts as a grid only when they read
+  like tiling: a couple of texts repeated across many fragments, the same text re-shown at one
+  spot over time (countdown/animation frames), or scattered single glyphs. Secondary text omits
+  those grids instead of flattening a translated table or schedule into one synthetic line.
+  Wrapped lyric rows, CC-style dialogue blocks, and reconstructed single-line karaoke remain
+  eligible for display.
 - The resolved text is stored in `mpvClient.currentSecondarySubText` before it is broadcast. The
  overlay, mining, timing tracker, and immersion statistics therefore consume the same secondary
  text when a readable source is available.
+- Simultaneous parsed cues use whitespace-insensitive identity, so ASS layers that vary only
+  between ordinary, hard, or ideographic spaces appear once.
+- Simultaneous ASS lines are flattened in top-to-bottom positioned order, falling back to their
+  authored source order when no usable position exists.
+- Half-size kana positioned directly above a same-timed kanji caption is treated as ASS
+  furigana. The parser omits it from published cues but retains hidden matching metadata so
+  mpv's raw live text can be reconciled without displaying or mining the reading. The
+  timing tracker (clipboard copy, recent-line mining) and immersion recorders run the same
+  reconciliation on the `sub-start`/`sub-end` sample, so they record what the overlay shows.
+- Broadcast-caption rows that spell one utterance across several same-timed positioned events
+  (same style, layer, and vertical band, stacked at most two text rows apart) are joined into
+  one cue with a single line break, so `preserveLineBreaks` treats them like an authored `\N`,
+  and the recorders above see the whole sentence. A row continues the one above it when that row
+  is a bare speaker label, ends without terminal punctuation, or leaves a ≪…≫ / ⸨…⸩ span open; a
+  lower row that opens its own label or span always starts a new cue, which keeps two speakers
+  sharing the screen on separate lines. The pass runs only on scripts that read as broadcast
+  captions (a meaningful share of events carry speaker labels or ≪…≫ / ⸨…⸩ spans) and only on
+  rows containing Japanese, because fansub typesetting stacks positioned rows for signs, chat
+  bubbles, and headlines where that punctuation convention does not hold.
+- Fragment-only ASS karaoke is reconstructed per style before publication. Explicit spaces
+  survive concatenation. Latin fragment typesetting with no literal spaces also recovers word
+  boundaries represented only by materially larger horizontal `\pos` or `\move` gaps within that
+  line. Unpositioned fragments stay compact instead of gaining guessed spaces between syllables.
+  Short runs qualify only when overlapping positioned events also show changing overrides or
+  repeated layer copies; an English or romaji style name alone never turns ordinary dialogue into
+  a lyric.
+- Recovered canonical ASS text remains active for the generated animation envelope. For
+  reconstructed lyric styles, the longest-lived active line wins over brief entrance and exit
+  fragments from the same style.
 - Media and `secondary-sid` changes clear the previous parsed state before refreshing the source;
  track-list changes refresh without discarding an unchanged source. Observed
  `secondary-sub-delay` changes retime the active parsed cue without rereading the file. If loading,
@@ -19,7 +19,7 @@ Read when: finding internal docs or checking verification status
 | Quality scorecard | `docs/knowledge-base/quality.md` | active | 2026-03-13 | quality grades and gaps |
 | Workflow index | `docs/workflow/README.md` | active | 2026-08-13 | execution map |
 | Planning guide | `docs/workflow/planning.md` | active | 2026-05-23 | lightweight vs execution plans |
-| Agent skills | `docs/workflow/agent-skills.md` | active | 2026-08-13 | repo-local workflow skill ownership |
+| Agent skills | `docs/workflow/agent-skills.md` | active | 2026-08-23 | repo-local workflow skill ownership |
 | Verification guide | `docs/workflow/verification.md` | active | 2026-08-13 | maintained verification lanes |
 | Release guide | `docs/RELEASING.md` | active | 2026-05-23 | release checklist |

@@ -3,7 +3,7 @@
 # Agent Skills

 Status: active
-Last verified: 2026-08-13
+Last verified: 2026-08-23
 Owner: Kyle Yasuda
 Read when: using, adding, or changing a repo-local agent workflow skill

@@ -12,6 +12,9 @@ Read when: using, adding, or changing a repo-local agent workflow skill
 - `.agents/skills/subminer-change-verification/`
  - Selects the cheapest sufficient repo-native verification lane.
  - Defers command ownership to `package.json` and `docs/workflow/verification.md`.
+- `.agents/skills/subminer-release/`
+  - Prepares, cuts, publishes, or repairs stable and prerelease releases.
+  - Defers release procedure and policy to `docs/RELEASING.md`.

 Repo-local workflows stay as standalone skills. Do not add plugin packaging, marketplace metadata, or compatibility shims unless the workflow is intentionally being distributed beyond this repository.

@@ -2,7 +2,7 @@
  "name": "subminer",
  "productName": "SubMiner",
  "desktopName": "SubMiner.desktop",
-  "version": "0.19.4-beta.2",
+  "version": "0.19.6",
  "description": "All-in-one sentence mining overlay with AnkiConnect and dictionary integration",
  "packageManager": "bun@1.3.5",
  "main": "dist/main-entry.js",
@@ -32,6 +32,7 @@
    "changelog:pr-check": "bun run scripts/build-changelog.ts pr-check",
    "changelog:release-notes": "bun run scripts/build-changelog.ts release-notes",
    "changelog:prerelease-notes": "bun run scripts/build-changelog.ts prerelease-notes",
+    "changelog:check-prerelease-notes": "bun run scripts/build-changelog.ts check-prerelease-notes",
    "format": "prettier --write .",
    "format:check": "prettier --check .",
    "format:src": "bash scripts/prettier-scope.sh --write",
@@ -86,7 +87,7 @@
    "app-builder-lib": "26.15.3",
    "brace-expansion": "5.0.9",
    "electron-builder-squirrel-windows": "26.15.3",
-    "fast-uri": "3.1.5",
+    "fast-uri": "3.1.6",
    "form-data": "4.0.6",
    "ip-address": "10.2.0",
    "js-yaml": "4.3.1",
@@ -6,38 +6,41 @@
 ### Added

 - Library Merge & Reassignment
-  - Duplicate library entries for the same show can now be merged: pick entries in "Select" mode and use "Merge Selected" to combine sessions, mined cards, and watch time onto one card.
-  - Episodes can be moved to a different library entry with a per-episode "→" button, fixing cases where a stray filename split off its own entry; manual assignments now survive later filename parsing, Jellyfin refreshes, and season repair.
-  - Exact AniList matches with compatible seasons now merge automatically, and likely (fuzzy) matches surface as a dismissible "Possible duplicate" suggestion instead of merging silently.
+  - Duplicate library cards for the same show can be combined: select entries in "Select" mode and use "Merge Selected" to combine their sessions, mined cards, and watch time onto one card.
+  - Episodes can be moved to a different entry with a per-episode "→" button, fixing stray files that split off their own entry; manual assignments now survive later filename parsing, Jellyfin refreshes, and season repair.
+  - Exact AniList matches with compatible seasons merge automatically, while likely (fuzzy) matches surface as a dismissible "Possible duplicate" suggestion instead of merging silently.

 - Duplicate Line Cleanup
-  - The Vocabulary tab's new **Duplicates** button scans a chosen time window for the repeated-line bursts described under Fixed below and collapses each burst to a single line once you confirm it; a matching `subminer stats cleanup --duplicate-lines` command (with `--dry-run` and `--lookback-days <n>`) is available from the terminal.
+  - The Vocabulary tab's new **Duplicates** button scans a chosen time window for the repeated-line bursts described under Fixed below and collapses each burst to a single line once confirmed.
+  - A matching `subminer stats cleanup --duplicate-lines` command (with `--dry-run` and `--lookback-days <n>`) is available from the terminal.
  - Only the affected subtitle lines and the vocabulary counts they inflated are touched; watch time and lines-seen totals are left as recorded.

 ### Fixed

 - Subtitle Duplication from Karaoke & Animated Signs
  - Typeset ASS karaoke and animated signs no longer flood the overlay, subtitle sidebar, immersion history, mined cards, or stats with repeated glyph fragments or per-frame duplicates; the complete authored line is recovered instead, without merging genuinely repeated dialogue or separately positioned signs.
-  - The secondary overlay now shares the same deduplication logic as the primary overlay, so layered animation text no longer appears multiple times there or in what gets mined.
-  - Vocabulary stats no longer count every animation frame of a karaoke opening as a separate line, which previously could push an OP lyric to the top of "Top Repeated Words."
+  - Fragmented karaoke now preserves the spaces the author placed between words instead of joining them together, and lyric transitions (including seeking into the middle of a line) resolve to the clean line instead of a stray entrance or exit frame.
+  - The secondary overlay shares the same deduplication logic as the primary overlay, including collapsing lines that differ only by whitespace or trailing punctuation, and sidebar navigation moves between clean lyric lines while keeping the right line selected.

 - Anki Media Generation
  - Sentence-audio generation no longer times out on slow network-mounted video files with many subtitle and font streams, and a failed extraction now reports a clear error instead of a raw `ENOENT`.
  - Mined audio and animated AVIF clips now capture the subtitle line you actually mined, instead of whatever line happened to be on screen once slow audio extraction finished.

 - Character Dictionary Performance & Notifications
-  - Character dictionary generation, merged rebuilds, and imports no longer freeze the app on large dictionaries, and cached results are reused across launches instead of regenerating character data and portraits every time.
+  - Character dictionary generation, merged rebuilds, and imports no longer freeze the app on large dictionaries, and cached results (including character portraits) are reused across launches instead of regenerating everything every time.
+  - Portraits also now display correctly if their cache finishes loading after subtitles have already started showing.
  - Desktop progress notifications, including on Linux AppImage installs, now update in place instead of flickering closed and reopening.

 - Overlay Reliability
-  - Overlay modals (settings, stats, etc.) now open promptly on the first shortcut press and appear above fullscreen mpv on macOS instead of switching Spaces or opening off-screen.
+  - Overlay modals (settings, stats, etc.) now open promptly on the first shortcut press, including on repeated sessions on Windows, and appear above fullscreen mpv on macOS instead of switching Spaces or opening off-screen.
+  - The macOS window-tracking helper is now built for macOS 12.0+, so the overlay attaches to mpv on older systems like Ventura instead of crashing and getting stuck on "Overlay loading."
  - The overlay no longer gets stuck on "Overlay loading" indefinitely if mpv's connection stalls; it now retries and shows an actionable error after 30 seconds.
-  - Fixed native Wayland drag-and-drop from file managers like Thunar, so subtitle and video files dropped on the overlay reach mpv.
-  - Fixed system-wide mouse lag on Windows caused by the overlay's click-through handling and repeated mpv window lookups.
+  - Fixed native Wayland drag-and-drop from file managers like Thunar, and fixed system-wide mouse lag on Windows caused by the overlay's click-through handling.

 - Stats Dashboard
  - Deletes, library merges, video moves, and AniList reassignments no longer freeze the stats dashboard or rebuild lifetime totals from scratch; large deletes that used to take minutes now finish in milliseconds.
-  - Vocabulary totals and charts now count all tracked vocabulary instead of just the first page, new-word history uses corrected daily rollups, calendar labels respect time zones west of UTC, and vocabulary cards refresh automatically after editing the word exclusion list.
+  - Vocabulary totals and charts now count all tracked vocabulary instead of just the first page, and new-word history uses corrected daily rollups.
+  - Calendar labels respect time zones west of UTC, and vocabulary cards refresh automatically after editing the word exclusion list (with a Retry option if a load fails).

 - Linux Launcher Thumbnails
  - Fixed missing MKV thumbnails in the Linux rofi picker when the system thumbnailer only registers legacy Matroska MIME aliases.
@@ -0,0 +1,51 @@
+## Highlights
+### Added
+- **Pre-Mining Timing Review**:
+  - Optional review step before creating word, sentence, or audio cards, with a speech-focused waveform that filters out steady background noise so dialogue is easy to spot.
+  - The clip end automatically snaps back to where dialogue actually ends, since subtitles often linger after speech stops.
+  - Drag or use the keyboard to adjust clip boundaries, and preview audio with a sweeping playhead that plays to the true end even on high-latency outputs like Bluetooth headphones.
+  - Pull extra previous or next subtitle lines onto the card with `P`/`N` (or the Prev/Next steppers); a live preview shows exactly what the card will contain.
+  - You can cancel and still keep the card without media, and the review can be toggled on or off for the session.
+- **Senren Note Type Support**:
+  - Enable `ankiConnect.isSenren` to merge duplicate mined cards using Senren's scene-switching markup, combining sentence, furigana, audio, picture, and misc-info fields.
+  - Supports the same auto/manual/disabled grouping modes as Kiku, including the manual merge modal. Senren and Kiku are mutually exclusive, so only one can be enabled at a time.
+
+### Changed
+- **Remote Streaming Mining**: Mining a card from a remote stream (Jellyfin and other HTTP sources) now downloads the clip window once and reuses it for the timing review waveform, audio preview, audio extraction, and screenshot, instead of re-fetching the stream for every step. No action needed; the temporary download is cleaned up automatically after ten minutes of inactivity.
+- **TsukiHime Release Picker**: The Japanese and secondary-language tabs now filter releases down to ones that actually carry subtitles for that language, and tell you when none do.
+
+### Fixed
+- **Broadcast Caption Accuracy**:
+  - Japanese caption tracks split across two positioned lines (e.g. Crunchyroll) now merge into one, so mined sentences, the sidebar, and line-break settings treat them as a single line; lines from different speakers or sound effects still stay separate.
+  - Mining from the overlay no longer picks up a leftover line from the previous caption, so the mined sentence and clip timing match what's actually on screen.
+  - Copying or mining subtitles no longer includes the separate furigana line that some broadcast subtitle files place above kanji.
+- **Multi-line Copy After Seeking**: Selecting multiple subtitle lines to copy or mine now selects backward in timeline order after a seek, rather than in playback encounter order.
+- **Overlay Stability**:
+  - On Hyprland, opening a modal (timing review, Jimaku, session help, and others) over fullscreen mpv no longer makes the overlay flicker while the modal loads.
+  - Switching secondary subtitle tracks no longer causes mpv's native secondary subtitles to flash on screen.
+- **Anki Update Notifications**: Switching notification settings to on-screen display while a card update is still in progress now correctly dismisses the old overlay progress indicator.
+- **Jellyfin Subtitles**: Subtitle files now load with zero delay in mpv instead of Jellyfin inferring and applying a sync offset.
+
+## What's Changed
+
+- feat(anki): add media timing review before card creation by @ksyasuda in #203
+- fix(jellyfin): stop inferring subtitle delays by @ksyasuda in #227
+- feat(anki): support Senren scene-switching field grouping by @ksyasuda in #230
+- fix(mining): copy multi-line subtitles backward from current line by @ksyasuda in #231
+- fix(subtitles): keep native secondary subtitles hidden by @ksyasuda in #232
+- fix(subtitles): drop ASS furigana from recorded cues by @ksyasuda in #233
+- fix(subtitles): merge wrapped positioned caption rows by @ksyasuda in #234
+- fix(tsukihime): filter releases by subtitle language by @ksyasuda in #235
+
+## Installation
+
+See the README and docs/installation guide for full setup steps.
+
+## Assets
+
+- Linux: `SubMiner.AppImage`
+- macOS: `SubMiner-*.dmg` and `SubMiner-*.zip`
+- Windows: `SubMiner-*.exe` and `SubMiner-*-win.zip`
+- Optional extras: `subminer-assets.tar.gz` and the `subminer` launcher
+
+Note: the `subminer` wrapper script uses Bun (`#!/usr/bin/env bun`), so `bun` must be installed and on `PATH`.
@@ -1,4 +1,5 @@
 import assert from 'node:assert/strict';
+import { execFileSync } from 'node:child_process';
 import fs from 'node:fs';
 import path from 'node:path';
 import test from 'node:test';
@@ -43,14 +44,22 @@ function fragmentTypesInPrompt(input: string): string[] {
    .map((line) => line.slice('type: '.length).trim());
 }

-function assertReleaseNotesPromptRequestsNestedBullets(input: string): void {
-  assert.match(input, /In MODE: release-notes, use short top-level change bullets/);
-  assert.match(input, /Nested bullets should cover the change, user benefit, and any user action/);
-  assert.match(input, /Do not require the exact nested labels/);
+function assertPromptRequestsNestedBullets(input: string): void {
+  assert.match(input, /In both modes, split every item into one nested bullet per distinct change/);
+  assert.match(input, /Never stack several distinct changes into one long paragraph-shaped bullet/);
  assert.match(input, /Keep nested bullets short, concrete, and readable by non-technical users/);
  assert.match(input, /Avoid paragraph-style release-note bullets/);
 }

+function assertReleaseNotesPromptRequestsNestedBullets(input: string): void {
+  assertPromptRequestsNestedBullets(input);
+  assert.match(
+    input,
+    /In MODE: release-notes, nested bullets should also cover user benefit and any user action/,
+  );
+  assert.match(input, /Do not require the exact nested labels/);
+}
+
 function defaultPolishedBody(input: string): string {
  const mode = modeFromPrompt(input);
  const types = fragmentTypesInPrompt(input);
@@ -445,6 +454,7 @@ test('writeChangelogArtifacts prompts Claude to summarize the final stable outco
        prompt,
        /Multiple fixes within the same prerelease cycle should collapse into one current-state bullet/,
      );
+      assertPromptRequestsNestedBullets(prompt);
    }

    const releaseNotesPrompt = stub.calls.find(
@@ -583,7 +593,7 @@ test('writePrereleaseNotesForVersion writes cumulative beta notes without mutati
    const outputPath = writePrereleaseNotesForVersion({
      cwd: projectRoot,
      version: '0.11.3-beta.1',
-      deps: { runClaude: stub.runClaude },
+      deps: { runClaude: stub.runClaude, listPrereleaseTags: () => [] },
    });

    assert.equal(outputPath, path.join(projectRoot, 'release', 'prerelease-notes.md'));
@@ -605,7 +615,8 @@ test('writePrereleaseNotesForVersion writes cumulative beta notes without mutati

    const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
    assert.match(prereleaseNotes, /^> This is a prerelease build for testing\./m);
-    assert.match(prereleaseNotes, /<!-- prerelease-base-version: 0\.11\.3 -->/);
+    assert.match(prereleaseNotes, /<!-- prerelease-version: 0\.11\.3-beta\.1 -->/);
+    assert.doesNotMatch(prereleaseNotes, /## Changes since /);
    assert.match(prereleaseNotes, /## Highlights\n### Added\n- Polished: added entry\./);
    assert.match(prereleaseNotes, /### Fixed\n- Polished: fixed entry\./);
    assert.match(prereleaseNotes, /## Installation\n\nSee the README and docs\/installation guide/);
@@ -668,7 +679,7 @@ test('writePrereleaseNotesForVersion reuses existing prerelease notes when addin
    const outputPath = writePrereleaseNotesForVersion({
      cwd: projectRoot,
      version: '0.11.3-beta.2',
-      deps: { runClaude: stub.runClaude },
+      deps: { runClaude: stub.runClaude, listPrereleaseTags: () => [] },
    });

    assert.equal(stub.calls.length, 1, 'prerelease should issue exactly one Claude call');
@@ -723,7 +734,7 @@ test('writePrereleaseNotesForVersion ignores unmarked prerelease notes from an o
    const outputPath = writePrereleaseNotesForVersion({
      cwd: projectRoot,
      version: '0.17.0-beta.1',
-      deps: { runClaude: stub.runClaude },
+      deps: { runClaude: stub.runClaude, listPrereleaseTags: () => [] },
    });

    assert.equal(stub.calls.length, 1, 'prerelease should issue exactly one Claude call');
@@ -790,7 +801,7 @@ test('writePrereleaseNotesForVersion prompts Claude to revise stale prerelease b
    writePrereleaseNotesForVersion({
      cwd: projectRoot,
      version: '0.12.0-beta.2',
-      deps: { runClaude: stub.runClaude },
+      deps: { runClaude: stub.runClaude, listPrereleaseTags: () => [] },
    });

    assert.equal(stub.calls.length, 1, 'prerelease should issue exactly one Claude call');
@@ -830,7 +841,7 @@ test('writePrereleaseNotesForVersion supports rc prereleases', async () => {
    const outputPath = writePrereleaseNotesForVersion({
      cwd: projectRoot,
      version: '0.11.3-rc.1',
-      deps: { runClaude: stub.runClaude },
+      deps: { runClaude: stub.runClaude, listPrereleaseTags: () => [] },
    });

    const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
@@ -1447,3 +1458,373 @@ test('writeChangelogArtifacts strips <details> blocks from release notes when re
    fs.rmSync(workspace, { recursive: true, force: true });
  }
 });
+
+test('selectPreviousPrereleaseTag orders betas before rcs and filters other base versions', async () => {
+  const { selectPreviousPrereleaseTag } = await loadModule();
+
+  const tags = [
+    'v0.19.4-beta.1',
+    'v0.19.4-beta.3',
+    'v0.19.4-beta.2',
+    'v0.19.3-beta.9',
+    'v0.19.4-rc.1',
+    'not-a-tag',
+  ];
+
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-beta.1'), null);
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-beta.2'), 'v0.19.4-beta.1');
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-beta.4'), 'v0.19.4-beta.3');
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-rc.1'), 'v0.19.4-beta.3');
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-rc.2'), 'v0.19.4-rc.1');
+  // Regenerating notes for an already-tagged version must not pick itself.
+  assert.equal(selectPreviousPrereleaseTag(tags, '0.19.4-beta.3'), 'v0.19.4-beta.2');
+  assert.equal(selectPreviousPrereleaseTag(['v0.19.3-beta.1'], '0.19.4-beta.2'), null);
+});
+
+test('writePrereleaseNotesForVersion adds a delta section generated from fragment diffs', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-delta-section');
+  const projectRoot = path.join(workspace, 'SubMiner');
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.12.0-beta.2' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', '001.md'),
+    ['type: fixed', 'area: overlay', '', '- Fixed overlay focus and macOS helper.'].join('\n'),
+    'utf8',
+  );
+
+  try {
+    const stub = recordingRunClaude((input) =>
+      input.includes('MODIFIED FRAGMENT')
+        ? '- Fixed the macOS helper deployment target for older systems.'
+        : '### Fixed\n- Overlay: cumulative fixed entry.',
+    );
+    const outputPath = writePrereleaseNotesForVersion({
+      cwd: projectRoot,
+      version: '0.12.0-beta.2',
+      deps: {
+        runClaude: stub.runClaude,
+        listPrereleaseTags: () => ['v0.12.0-beta.1'],
+        resolveFragmentDelta: (_cwd, previousTag) => {
+          assert.equal(previousTag, 'v0.12.0-beta.1');
+          return [
+            {
+              path: 'changes/002.md',
+              status: 'added',
+              after: 'type: fixed\narea: macos\n\n- Fixed helper deployment target.',
+            },
+            {
+              path: 'changes/001.md',
+              status: 'modified',
+              before: '- Fixed overlay focus.',
+              after: '- Fixed overlay focus and macOS helper.',
+            },
+            {
+              path: 'changes/003.md',
+              status: 'deleted',
+              before: 'type: added\narea: stats\n\n- Reverted experimental stats view.',
+            },
+          ];
+        },
+      },
+    });
+
+    assert.equal(stub.calls.length, 2, 'delta and cumulative polish are separate Claude calls');
+    const deltaPrompt = stub.calls[0]!.input;
+    assert.match(deltaPrompt, /ADDED FRAGMENT changes\/002\.md/);
+    assert.match(deltaPrompt, /MODIFIED FRAGMENT changes\/001\.md/);
+    assert.match(deltaPrompt, /BEFORE:\n- Fixed overlay focus\./);
+    assert.match(deltaPrompt, /AFTER:\n- Fixed overlay focus and macOS helper\./);
+    assert.match(deltaPrompt, /DELETED FRAGMENT changes\/003\.md/);
+    assert.match(deltaPrompt, /If the edit is editorial/);
+    assert.match(deltaPrompt, /removed or reverted/);
+    assert.match(deltaPrompt, /No user-facing changes since v0\.12\.0-beta\.1\./);
+    assert.equal(modeFromPrompt(stub.calls[1]!.input), 'release-notes');
+
+    const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
+    assert.match(
+      prereleaseNotes,
+      /<!-- prerelease-version: 0\.12\.0-beta\.2; since: v0\.12\.0-beta\.1 -->/,
+    );
+    const deltaIndex = prereleaseNotes.indexOf('## Changes since v0.12.0-beta.1');
+    const highlightsIndex = prereleaseNotes.indexOf('## Highlights');
+    assert.ok(deltaIndex !== -1, 'delta section heading should be present');
+    assert.ok(deltaIndex < highlightsIndex, 'delta section should precede Highlights');
+    assert.match(prereleaseNotes, /- Fixed the macOS helper deployment target for older systems\./);
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
+test('writePrereleaseNotesForVersion renders a fallback delta line when no fragments changed', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-empty-delta');
+  const projectRoot = path.join(workspace, 'SubMiner');
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.12.0-beta.3' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', '001.md'),
+    ['type: fixed', 'area: overlay', '', '- Fixed overlay focus.'].join('\n'),
+    'utf8',
+  );
+
+  try {
+    const stub = defaultStubClaude();
+    const outputPath = writePrereleaseNotesForVersion({
+      cwd: projectRoot,
+      version: '0.12.0-beta.3',
+      deps: {
+        runClaude: stub.runClaude,
+        listPrereleaseTags: () => ['v0.12.0-beta.1', 'v0.12.0-beta.2'],
+        resolveFragmentDelta: () => [],
+      },
+    });
+
+    assert.equal(stub.calls.length, 1, 'empty delta must not spend a Claude call');
+    const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
+    assert.match(
+      prereleaseNotes,
+      /## Changes since v0\.12\.0-beta\.2\n\n- No changelog fragment changes since v0\.12\.0-beta\.2; this build contains packaging or internal-only updates\./,
+    );
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
+test('writePrereleaseNotesForVersion rejects non-bullet delta output from Claude', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-delta-invalid-output');
+  const projectRoot = path.join(workspace, 'SubMiner');
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.12.0-beta.2' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', '001.md'),
+    ['type: fixed', 'area: overlay', '', '- Fixed overlay focus.'].join('\n'),
+    'utf8',
+  );
+
+  try {
+    const stub = recordingRunClaude(() => 'Here are the changes:\n- One change.');
+    assert.throws(
+      () =>
+        writePrereleaseNotesForVersion({
+          cwd: projectRoot,
+          version: '0.12.0-beta.2',
+          deps: {
+            runClaude: stub.runClaude,
+            listPrereleaseTags: () => ['v0.12.0-beta.1'],
+            resolveFragmentDelta: () => [
+              { path: 'changes/001.md', status: 'added', after: '- Fixed overlay focus.' },
+            ],
+          },
+        }),
+      /delta output must contain only Markdown bullets/,
+    );
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
+test('writePrereleaseNotesForVersion strips the stale delta section from the reused baseline', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-reuse-strips-delta');
+  const projectRoot = path.join(workspace, 'SubMiner');
+  const existingNotes = [
+    '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
+    '',
+    '<!-- prerelease-version: 0.12.0-beta.2; since: v0.12.0-beta.1 -->',
+    '',
+    '## Changes since v0.12.0-beta.1',
+    '',
+    '- Stale beta-to-beta delta bullet.',
+    '',
+    '## Highlights',
+    '### Added',
+    '- Overlay: Previous beta entry.',
+    '',
+    '## Installation',
+    '',
+    'See the README and docs/installation guide for full setup steps.',
+    '',
+  ].join('\n');
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.mkdirSync(path.join(projectRoot, 'release'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.12.0-beta.3' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(path.join(projectRoot, 'release', 'prerelease-notes.md'), existingNotes, 'utf8');
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', '001.md'),
+    ['type: added', 'area: overlay', '', '- Added overlay coverage.'].join('\n'),
+    'utf8',
+  );
+
+  try {
+    const stub = defaultStubClaude();
+    writePrereleaseNotesForVersion({
+      cwd: projectRoot,
+      version: '0.12.0-beta.3',
+      deps: {
+        runClaude: stub.runClaude,
+        listPrereleaseTags: () => [],
+        resolveFragmentDelta: () => [],
+      },
+    });
+
+    assert.equal(stub.calls.length, 1);
+    const prompt = stub.calls[0]!.input;
+    assert.match(prompt, /EXISTING PRERELEASE NOTES/);
+    assert.match(prompt, /Overlay: Previous beta entry\./);
+    assert.doesNotMatch(prompt, /Stale beta-to-beta delta bullet\./);
+    assert.doesNotMatch(prompt, /## Changes since /);
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
+test('verifyPrereleaseNotesMatchVersion accepts matching notes and rejects stale or legacy markers', async () => {
+  const { verifyPrereleaseNotesMatchVersion } = await loadModule();
+  const workspace = createWorkspace('verify-prerelease-notes');
+  const projectRoot = path.join(workspace, 'SubMiner');
+  const notesPath = path.join(projectRoot, 'release', 'prerelease-notes.md');
+
+  fs.mkdirSync(path.join(projectRoot, 'release'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.12.0-beta.2' }, null, 2),
+    'utf8',
+  );
+
+  try {
+    assert.throws(
+      () => verifyPrereleaseNotesMatchVersion({ cwd: projectRoot, version: '0.12.0-beta.2' }),
+      /Missing .*prerelease-notes\.md/,
+    );
+
+    fs.writeFileSync(
+      notesPath,
+      '<!-- prerelease-version: 0.12.0-beta.2; since: v0.12.0-beta.1 -->\n\n## Highlights\n',
+      'utf8',
+    );
+    verifyPrereleaseNotesMatchVersion({ cwd: projectRoot, version: '0.12.0-beta.2' });
+    verifyPrereleaseNotesMatchVersion({ cwd: projectRoot, version: 'v0.12.0-beta.2' });
+
+    fs.writeFileSync(
+      notesPath,
+      '<!-- prerelease-version: 0.12.0-beta.1 -->\n\n## Highlights\n',
+      'utf8',
+    );
+    assert.throws(
+      () => verifyPrereleaseNotesMatchVersion({ cwd: projectRoot, version: '0.12.0-beta.2' }),
+      /generated for 0\.12\.0-beta\.1 but this release is 0\.12\.0-beta\.2/,
+    );
+
+    fs.writeFileSync(
+      notesPath,
+      '<!-- prerelease-base-version: 0.12.0 -->\n\n## Highlights\n',
+      'utf8',
+    );
+    assert.throws(
+      () => verifyPrereleaseNotesMatchVersion({ cwd: projectRoot, version: '0.12.0-beta.2' }),
+      /missing or legacy prerelease-version marker/,
+    );
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
+test('default git tag listing and fragment delta resolution work against a real repository', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-git-defaults');
+  const projectRoot = path.join(workspace, 'SubMiner');
+  const git = (...args: string[]): void => {
+    execFileSync('git', args, { cwd: projectRoot, stdio: 'ignore' });
+  };
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.11.3-beta.1' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', 'kept.md'),
+    ['type: added', 'area: overlay', '', '- Kept change.'].join('\n'),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', 'edited.md'),
+    ['type: fixed', 'area: launcher', '', '- Original launcher fix.'].join('\n'),
+    'utf8',
+  );
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', 'removed.md'),
+    ['type: added', 'area: stats', '', '- Reverted stats change.'].join('\n'),
+    'utf8',
+  );
+
+  try {
+    git('init', '--quiet');
+    git('-c', 'user.email=test@example.com', '-c', 'user.name=Test', 'add', '.');
+    git('-c', 'user.email=test@example.com', '-c', 'user.name=Test', 'commit', '-m', 'beta.1');
+    git('tag', 'v0.11.3-beta.1');
+
+    fs.writeFileSync(
+      path.join(projectRoot, 'changes', 'edited.md'),
+      ['type: fixed', 'area: launcher', '', '- Broader launcher fix.'].join('\n'),
+      'utf8',
+    );
+    fs.rmSync(path.join(projectRoot, 'changes', 'removed.md'));
+    fs.writeFileSync(
+      path.join(projectRoot, 'changes', 'new.md'),
+      ['type: added', 'area: anki', '', '- New anki change.'].join('\n'),
+      'utf8',
+    );
+    fs.writeFileSync(
+      path.join(projectRoot, 'package.json'),
+      JSON.stringify({ name: 'subminer', version: '0.11.3-beta.2' }, null, 2),
+      'utf8',
+    );
+
+    const stub = recordingRunClaude((input) =>
+      input.includes('PREVIOUS_TAG:') ? '- Delta bullet.' : defaultPolishedBody(input),
+    );
+    writePrereleaseNotesForVersion({
+      cwd: projectRoot,
+      version: '0.11.3-beta.2',
+      deps: { runClaude: stub.runClaude },
+    });
+
+    assert.equal(stub.calls.length, 2);
+    const deltaPrompt = stub.calls[0]!.input;
+    assert.match(deltaPrompt, /PREVIOUS_TAG: v0\.11\.3-beta\.1/);
+    assert.match(deltaPrompt, /ADDED FRAGMENT changes\/new\.md/);
+    assert.match(deltaPrompt, /- New anki change\./);
+    assert.match(deltaPrompt, /MODIFIED FRAGMENT changes\/edited\.md/);
+    assert.match(deltaPrompt, /- Original launcher fix\./);
+    assert.match(deltaPrompt, /- Broader launcher fix\./);
+    assert.match(deltaPrompt, /DELETED FRAGMENT changes\/removed\.md/);
+    assert.match(deltaPrompt, /- Reverted stats change\./);
+    assert.doesNotMatch(deltaPrompt, /kept\.md/);
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
@@ -18,6 +18,15 @@ type Contribution = {
 // and the GitHub API.
 type ResolveContributions = (fragmentPaths: string[], cwd: string) => Contribution[];

+// One changelog fragment's change between the previous prerelease tag and the
+// working tree. `before` is the content at the tag, `after` the current content.
+export type FragmentDeltaEntry = {
+  path: string;
+  status: 'added' | 'modified' | 'deleted';
+  before?: string;
+  after?: string;
+};
+
 type ChangelogFsDeps = {
  existsSync?: (candidate: string) => boolean;
  mkdirSync?: (candidate: string, options: { recursive: true }) => void;
@@ -28,6 +37,8 @@ type ChangelogFsDeps = {
  log?: (message: string) => void;
  runClaude?: RunClaude;
  resolveContributions?: ResolveContributions;
+  listPrereleaseTags?: (cwd: string, baseVersion: string) => string[];
+  resolveFragmentDelta?: (cwd: string, previousTag: string) => FragmentDeltaEntry[];
 };

 type PolishMode = 'changelog' | 'release-notes';
@@ -103,16 +114,57 @@ function resolvePrereleaseBaseVersion(version: string): string {
  return match[1]!;
 }

-function renderPrereleaseBaseVersionMarker(version: string): string {
-  return `<!-- prerelease-base-version: ${resolvePrereleaseBaseVersion(version)} -->`;
+// The marker records which exact prerelease the committed notes were generated
+// for (and which prior tag the delta section compares against), so CI can
+// reject notes that were prepared for a different beta/RC.
+function renderPrereleaseVersionMarker(version: string, previousTag: string | null): string {
+  const since = previousTag ? `; since: ${previousTag}` : '';
+  return `<!-- prerelease-version: ${normalizeVersion(version)}${since} -->`;
 }

+export function extractPrereleaseVersionMarker(notes: string): string | null {
+  return (
+    /<!--\s*prerelease-version:\s*(\d+\.\d+\.\d+-(?:beta|rc)\.\d+)(?:;\s*since:\s*\S+)?\s*-->/u.exec(
+      notes,
+    )?.[1] ?? null
+  );
+}
+
+// Legacy marker written before the per-version marker existed. Still accepted
+// when deciding whether existing notes can seed the cumulative baseline.
 function extractPrereleaseBaseVersionMarker(notes: string): string | null {
+  const fullVersion = extractPrereleaseVersionMarker(notes);
+  if (fullVersion) {
+    return resolvePrereleaseBaseVersion(fullVersion);
+  }
  return /<!--\s*prerelease-base-version:\s*(\d+\.\d+\.\d+)\s*-->/u.exec(notes)?.[1] ?? null;
 }

+const DELTA_SECTION_HEADING_PREFIX = '## Changes since ';
+
+// Removes the previous run's "Changes since" section so the cumulative baseline
+// fed back to Claude never carries a stale beta-to-beta delta.
+function stripDeltaSection(notes: string): string {
+  const lines = notes.split(/\r?\n/);
+  const start = lines.findIndex((line) => line.startsWith(DELTA_SECTION_HEADING_PREFIX));
+  if (start === -1) {
+    return notes;
+  }
+  let end = lines.length;
+  for (let index = start + 1; index < lines.length; index += 1) {
+    if (lines[index]!.startsWith('## ')) {
+      end = index;
+      break;
+    }
+  }
+  return [...lines.slice(0, start), ...lines.slice(end)].join('\n');
+}
+
 function stripPrereleaseMetadata(notes: string): string {
-  return notes.replace(/<!--\s*prerelease-base-version:\s*\d+\.\d+\.\d+\s*-->\s*/u, '').trim();
+  return notes
+    .replace(/<!--\s*prerelease-version:[^>]*-->\s*/u, '')
+    .replace(/<!--\s*prerelease-base-version:\s*\d+\.\d+\.\d+\s*-->\s*/u, '')
+    .trim();
 }

 function resolveReusablePrereleaseNotes(notes: string, version: string): string | undefined {
@@ -120,7 +172,124 @@ function resolveReusablePrereleaseNotes(notes: string, version: string): string
  if (existingBaseVersion !== resolvePrereleaseBaseVersion(version)) {
    return undefined;
  }
-  return stripPrereleaseMetadata(notes);
+  return stripPrereleaseMetadata(stripDeltaSection(notes));
+}
+
+type ParsedPrereleaseTag = {
+  tag: string;
+  base: string;
+  channel: 'beta' | 'rc';
+  iteration: number;
+};
+
+function parsePrereleaseTag(tag: string): ParsedPrereleaseTag | null {
+  const match = /^v?(\d+\.\d+\.\d+)-(beta|rc)\.(\d+)$/u.exec(tag.trim());
+  if (!match) {
+    return null;
+  }
+  return {
+    tag: tag.trim(),
+    base: match[1]!,
+    channel: match[2] as 'beta' | 'rc',
+    iteration: Number.parseInt(match[3]!, 10),
+  };
+}
+
+// Semver prerelease order: every beta sorts before every rc, then numerically.
+function comparePrereleaseTags(a: ParsedPrereleaseTag, b: ParsedPrereleaseTag): number {
+  if (a.channel !== b.channel) {
+    return a.channel === 'beta' ? -1 : 1;
+  }
+  return a.iteration - b.iteration;
+}
+
+// Picks the newest prerelease tag for the same base version that strictly
+// precedes the version being released. Returns null for the first prerelease.
+export function selectPreviousPrereleaseTag(tags: string[], version: string): string | null {
+  const current = parsePrereleaseTag(normalizeVersion(version));
+  if (!current) {
+    return null;
+  }
+
+  const candidates = tags
+    .map(parsePrereleaseTag)
+    .filter((parsed): parsed is ParsedPrereleaseTag => parsed !== null)
+    .filter((parsed) => parsed.base === current.base)
+    .filter((parsed) => comparePrereleaseTags(parsed, current) < 0)
+    .sort(comparePrereleaseTags);
+
+  return candidates[candidates.length - 1]?.tag ?? null;
+}
+
+function defaultListPrereleaseTags(cwd: string, baseVersion: string): string[] {
+  return execFileSync('git', ['tag', '--list', `v${baseVersion}-beta.*`, `v${baseVersion}-rc.*`], {
+    cwd,
+    encoding: 'utf8',
+  })
+    .split(/\r?\n/)
+    .map((line) => line.trim())
+    .filter(Boolean);
+}
+
+// Diffs changes/*.md between the previous prerelease tag and the working tree.
+// Renamed fragments are treated as modifications of the new path.
+//
+// Like every other path in this script, git paths are resolved against `cwd`,
+// which is the project root and also the repository root. Callers that point
+// `cwd` elsewhere already fail earlier and loudly, when package.json and
+// changes/ come back missing.
+function defaultResolveFragmentDelta(cwd: string, previousTag: string): FragmentDeltaEntry[] {
+  const output = execFileSync(
+    'git',
+    ['diff', '--name-status', '--find-renames', previousTag, '--', 'changes'],
+    { cwd, encoding: 'utf8' },
+  );
+  const showAtTag = (fragmentPath: string): string =>
+    execFileSync('git', ['show', `${previousTag}:${fragmentPath}`], { cwd, encoding: 'utf8' });
+  const readCurrent = (fragmentPath: string): string =>
+    fs.readFileSync(path.join(cwd, fragmentPath), 'utf8');
+
+  const entries: FragmentDeltaEntry[] = [];
+  for (const line of output.split(/\r?\n/)) {
+    if (!line.trim()) {
+      continue;
+    }
+    const [status = '', ...paths] = line.split('\t');
+    const oldPath = paths[0] ?? '';
+    const newPath = paths[paths.length - 1] ?? '';
+    if (!isFragmentPath(newPath) && !isFragmentPath(oldPath)) {
+      continue;
+    }
+
+    if (status.startsWith('A')) {
+      entries.push({ path: newPath, status: 'added', after: readCurrent(newPath) });
+    } else if (status.startsWith('D')) {
+      entries.push({ path: oldPath, status: 'deleted', before: showAtTag(oldPath) });
+    } else if (status.startsWith('M') || status.startsWith('R')) {
+      entries.push({
+        path: newPath,
+        status: 'modified',
+        before: showAtTag(oldPath),
+        after: readCurrent(newPath),
+      });
+    }
+  }
+
+  // git diff misses fragments that exist only in the working tree; treat
+  // untracked fragments as additions so a pre-commit run still sees them.
+  const untracked = execFileSync(
+    'git',
+    ['ls-files', '--others', '--exclude-standard', '--', 'changes'],
+    { cwd, encoding: 'utf8' },
+  )
+    .split(/\r?\n/)
+    .map((line) => line.trim())
+    .filter((candidate) => candidate && isFragmentPath(candidate));
+  for (const fragmentPath of untracked) {
+    entries.push({ path: fragmentPath, status: 'added', after: readCurrent(fragmentPath) });
+  }
+
+  return entries;
 }

 function verifyRequestedVersionMatchesPackageVersion(
@@ -311,10 +480,15 @@ You will receive a list of FRAGMENT entries below. Each fragment has metadata (t
   - Be merged with related bullets when possible. If five fragments all touch Windows overlay z-order/focus/restore, write one or two bullets that summarize the overall improvement instead of five.
   - Drop bullets that only describe PR housekeeping, CodeRabbit follow-ups, or test-only changes that don't affect users.
   - Preserve the substance of breaking changes that remain breaking after applying the Release Outcome Rules. Do not soften or omit them.
-5. In MODE: changelog, each item may be a conventional single-level bullet, e.g. "- Playlist Browser: Adds faster saved-show browsing."
-6. In MODE: release-notes, use short top-level change bullets with two or three nested bullets when an item needs explanation.
-   Nested bullets should cover the change, user benefit, and any user action or compatibility note when useful. Do not require the exact nested labels; natural phrasing is fine. Omit the action bullet when no action is needed.
+5. In both modes, split every item into one nested bullet per distinct change. Write a short bold name on the top-level bullet, then indent the details two spaces:
+   - **Playlist Browser**:
+     - Saved shows now open without rescanning the library.
+     - The picker remembers the last folder you browsed between launches.
+   Each nested bullet covers exactly one change, behavior, or user-visible outcome. Never stack several distinct changes into one long paragraph-shaped bullet.
+   Aim for two to five nested bullets per item. When an item genuinely has only one thing to say, put it inline on the top-level bullet ("- **Playlist Browser**: Saved shows now open without rescanning the library.") instead of emitting a single nested bullet.
   Keep nested bullets short, concrete, and readable by non-technical users. Avoid paragraph-style release-note bullets.
+   Bullets inside the Internal section may stay single-level.
+6. In MODE: release-notes, nested bullets should also cover user benefit and any user action or compatibility note when useful. Do not require the exact nested labels; natural phrasing is fine. Omit the action bullet when no action is needed.
 7. Do not invent features. Every bullet must be grounded in the input fragments.
 8. Do not include the version heading (## v...) — that wrapper is added by the caller.

@@ -615,7 +789,7 @@ function polishFragmentsWithClaude(
    ? [
        '## Existing Prerelease Notes',
        '',
-        'The input includes EXISTING PRERELEASE NOTES before the fragment list. Existing prerelease notes are a baseline, not an immutable changelog. Reuse reviewed highlight bullets when they still describe the current outcome, but replace stale beta or RC wording when new fragments supersede it. Merge in only new or changed fragment material, and deduplicate instead of restating existing bullets. Output only the final highlights body using the section headings above; do not include the prerelease disclaimer, Installation, or Assets sections.',
+        'The input includes EXISTING PRERELEASE NOTES before the fragment list. Existing prerelease notes are a baseline, not an immutable changelog. Reuse reviewed highlight bullets when they still describe the current outcome, but replace stale beta or RC wording when new fragments supersede it. Merge in only new or changed fragment material, and deduplicate instead of restating existing bullets. Output only the final highlights body using the section headings above; do not include the prerelease disclaimer, any "Changes since" section, or the Installation or Assets sections.',
        '',
      ].join('\n')
    : '';
@@ -627,6 +801,75 @@ function polishFragmentsWithClaude(
  return validatePolishedOutput(output, mode, hasInternalFragments);
 }

+const DELTA_PROMPT_INSTRUCTIONS = `You are writing the "changes since the previous prerelease" section of a prerelease notes file for SubMiner, an Electron app for Japanese sentence mining.
+
+You will receive changelog fragment diffs between the previous prerelease tag and the current build. Fragments are engineer-written release-note sources; a fragment diff is a proxy for what changed, not proof of a behavior change.
+
+Rules:
+
+1. Output Markdown bullets ONLY. No headings, no preamble, no commentary. Every line must be a top-level "- " bullet or an indented nested bullet.
+2. Describe only what changed for users between the two prerelease builds, in user-facing language. Drop implementation jargon, file paths, and PR numbers.
+3. ADDED fragments describe changes that are new in this build; summarize them.
+4. MODIFIED fragments include BEFORE and AFTER content. Describe only the behavioral difference between them. If the edit is editorial (rewording, deduplication, reformatting, reconciling stale phrasing) with no user-visible behavior change, omit it entirely.
+5. DELETED fragments mean the described change was removed or reverted before this build; say so explicitly.
+6. Keep bullets short and concrete. Use nested bullets sparingly.
+7. Do not invent changes. Every bullet must be grounded in the diffs.
+8. If no bullet survives rules 2-5, output exactly this single line:
+   - No user-facing changes since PREVIOUS_TAG.
+
+The input begins below.
+
+`;
+
+function serializeFragmentDeltaForPrompt(
+  delta: FragmentDeltaEntry[],
+  version: string,
+  previousTag: string,
+): string {
+  const header = [`VERSION: ${version}`, `PREVIOUS_TAG: ${previousTag}`];
+  const blocks = delta.map((entry) => {
+    if (entry.status === 'added') {
+      return [`ADDED FRAGMENT ${entry.path}`, entry.after ?? ''].join('\n');
+    }
+    if (entry.status === 'deleted') {
+      return [`DELETED FRAGMENT ${entry.path}`, entry.before ?? ''].join('\n');
+    }
+    return [
+      `MODIFIED FRAGMENT ${entry.path}`,
+      'BEFORE:',
+      entry.before ?? '',
+      'AFTER:',
+      entry.after ?? '',
+    ].join('\n');
+  });
+  return [...header, '', ...blocks].join('\n\n');
+}
+
+function validateDeltaOutput(output: string): string {
+  const trimmed = output.trim();
+  if (!trimmed) {
+    throw new Error('claude returned empty output for the prerelease delta section.');
+  }
+  const invalidLine = trimmed.split(/\r?\n/).find((line) => line.trim() && !/^\s*- /.test(line));
+  if (invalidLine !== undefined) {
+    throw new Error(
+      `claude delta output must contain only Markdown bullets. Offending line:\n${invalidLine}`,
+    );
+  }
+  return trimmed;
+}
+
+function buildDeltaSectionWithClaude(
+  delta: FragmentDeltaEntry[],
+  options: { version: string; previousTag: string; deps?: ChangelogFsDeps },
+): string {
+  const runClaude = options.deps?.runClaude ?? defaultRunClaude;
+  const prompt =
+    DELTA_PROMPT_INSTRUCTIONS.replace('PREVIOUS_TAG', options.previousTag) +
+    serializeFragmentDeltaForPrompt(delta, options.version, options.previousTag);
+  return validateDeltaOutput(runClaude(prompt, CLAUDE_CLI_ARGS));
+}
+
 function stripDetailsBlocks(body: string): string {
  return body.replace(/<details>[\s\S]*?<\/details>\s*/gm, '').trim();
 }
@@ -709,15 +952,18 @@ function renderReleaseNotes(
    contributions?: Contribution[];
    contributorSections?: string[];
    metadata?: string[];
+    deltaSection?: string[];
  },
 ): string {
  const prefix = options?.disclaimer ? [options.disclaimer, ''] : [];
  const metadata = options?.metadata?.length ? [...options.metadata, ''] : [];
+  const deltaSection = options?.deltaSection?.length ? [...options.deltaSection, ''] : [];
  const contributorSections =
    options?.contributorSections ?? renderContributorsSections(options?.contributions ?? []);
  return [
    ...prefix,
    ...metadata,
+    ...deltaSection,
    '## Highlights',
    changes,
    '',
@@ -748,6 +994,7 @@ function writeReleaseNotesFile(
    contributions?: Contribution[];
    contributorSections?: string[];
    metadata?: string[];
+    deltaSection?: string[];
  },
 ): string {
  const mkdirSync = deps?.mkdirSync ?? fs.mkdirSync;
@@ -1079,6 +1326,26 @@ export function writePrereleaseNotesForVersion(options?: ChangelogOptions): stri
    throw new Error('No changelog fragments found in changes/.');
  }

+  const listPrereleaseTags = options?.deps?.listPrereleaseTags ?? defaultListPrereleaseTags;
+  const previousTag = selectPreviousPrereleaseTag(
+    listPrereleaseTags(cwd, resolvePrereleaseBaseVersion(version)),
+    version,
+  );
+
+  // Later betas/RCs get a "Changes since <previous tag>" section on top of the
+  // cumulative Highlights, generated from the fragment diff between the
+  // previous prerelease tag and the working tree.
+  let deltaSection: string[] = [];
+  if (previousTag) {
+    const resolveFragmentDelta = options?.deps?.resolveFragmentDelta ?? defaultResolveFragmentDelta;
+    const delta = resolveFragmentDelta(cwd, previousTag);
+    const deltaBody =
+      delta.length === 0
+        ? `- No changelog fragment changes since ${previousTag}; this build contains packaging or internal-only updates.`
+        : buildDeltaSectionWithClaude(delta, { version, previousTag, deps: options?.deps });
+    deltaSection = [`${DELTA_SECTION_HEADING_PREFIX}${previousTag}`, '', deltaBody];
+  }
+
  const prereleaseNotesPath = path.join(cwd, PRERELEASE_NOTES_PATH);
  const existingReleaseNotes = existsSync(prereleaseNotesPath)
    ? resolveReusablePrereleaseNotes(readFileSync(prereleaseNotesPath, 'utf8'), version)
@@ -1095,10 +1362,41 @@ export function writePrereleaseNotesForVersion(options?: ChangelogOptions): stri
      '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
    outputPath: PRERELEASE_NOTES_PATH,
    contributions,
-    metadata: [renderPrereleaseBaseVersionMarker(version)],
+    metadata: [renderPrereleaseVersionMarker(version, previousTag)],
+    deltaSection,
  });
 }

+// CI gate: the committed prerelease notes must carry a marker generated for
+// exactly the version being tagged, so stale beta.N-1 notes can't ship.
+export function verifyPrereleaseNotesMatchVersion(options?: ChangelogOptions): void {
+  verifyRequestedVersionMatchesPackageVersion(options ?? {});
+
+  const cwd = options?.cwd ?? process.cwd();
+  const existsSync = options?.deps?.existsSync ?? fs.existsSync;
+  const readFileSync = options?.deps?.readFileSync ?? fs.readFileSync;
+  const version = resolveVersion(options ?? {});
+  if (!isSupportedPrereleaseVersion(version)) {
+    throw new Error(
+      `Unsupported prerelease version (${version}). Expected x.y.z-beta.N or x.y.z-rc.N.`,
+    );
+  }
+
+  const prereleaseNotesPath = path.join(cwd, PRERELEASE_NOTES_PATH);
+  if (!existsSync(prereleaseNotesPath)) {
+    throw new Error(
+      `Missing ${prereleaseNotesPath}. Run 'bun run changelog:prerelease-notes --version ${version}' and commit the file before tagging.`,
+    );
+  }
+
+  const markerVersion = extractPrereleaseVersionMarker(readFileSync(prereleaseNotesPath, 'utf8'));
+  if (markerVersion !== version) {
+    throw new Error(
+      `release/prerelease-notes.md was generated for ${markerVersion ?? 'an unknown version (missing or legacy prerelease-version marker)'} but this release is ${version}. Rerun 'bun run changelog:prerelease-notes --version ${version}' and commit the result.`,
+    );
+  }
+}
+
 function parseCliArgs(argv: string[]): {
  baseRef?: string;
  cwd?: string;
@@ -1206,6 +1504,11 @@ function main(): void {
    return;
  }

+  if (command === 'check-prerelease-notes') {
+    verifyPrereleaseNotesMatchVersion(options);
+    return;
+  }
+
  if (command === 'docs') {
    generateDocsChangelog(options);
    return;
@@ -1,54 +0,0 @@
-#!/bin/bash
-# Build macOS window tracking helper binary
-
-set -e
-
-SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
-SWIFT_SOURCE="$SCRIPT_DIR/get-mpv-window-macos.swift"
-OUTPUT_DIR="$SCRIPT_DIR/../dist/scripts"
-OUTPUT_BINARY="$OUTPUT_DIR/get-mpv-window-macos"
-OUTPUT_SOURCE_COPY="$OUTPUT_DIR/get-mpv-window-macos.swift"
-
-fallback_to_source() {
-  echo "Falling back to source fallback: $OUTPUT_SOURCE_COPY"
-  mkdir -p "$OUTPUT_DIR"
-  cp "$SWIFT_SOURCE" "$OUTPUT_SOURCE_COPY"
-}
-
-build_swift_helper() {
-  echo "Compiling macOS window tracking helper..."
-  if ! command -v swiftc >/dev/null 2>&1; then
-    echo "swiftc not found in PATH; skipping compilation."
-    return 1
-  fi
-
-  if ! swiftc -O "$SWIFT_SOURCE" -o "$OUTPUT_BINARY"; then
-    return 1
-  fi
-
-  chmod +x "$OUTPUT_BINARY"
-  echo "✓ Built $OUTPUT_BINARY"
-  return 0
-}
-
-# Optional skip flag for non-macOS CI/dev environments
-if [[ "${SUBMINER_SKIP_MACOS_HELPER_BUILD:-}" == "1" ]]; then
-  echo "Skipping macOS helper build (SUBMINER_SKIP_MACOS_HELPER_BUILD=1)"
-  fallback_to_source
-  exit 0
-fi
-
-# Only build on macOS
-if [[ "$(uname)" != "Darwin" ]]; then
-  echo "Skipping macOS helper build (not on macOS)"
-  fallback_to_source
-  exit 0
-fi
-
-# Create output directory
-mkdir -p "$OUTPUT_DIR"
-
-# Compile Swift script to binary, fallback to source if unavailable or compilation fails
-if ! build_swift_helper; then
-  fallback_to_source
-fi
@@ -0,0 +1,17 @@
+#!/usr/bin/env bash
+
+set -euo pipefail
+
+FILE="${1:-}"
+
+if [[ ! -f "$FILE" ]]; then
+	printf 'Not a file: %s\n' "${FILE:-<missing>}" >&2
+	exit 1
+fi
+
+if ! mpv --no-config --no-terminal --msg-level=all=no --vo=null --ao=null --frames=1 -- "$FILE"; then
+	printf 'Not playable by mpv: %s\n' "$FILE" >&2
+	exit 1
+fi
+
+exec subminer app --dev --launch-mpv "$FILE"
@@ -1,4 +1,5 @@
 import fs from 'node:fs';
+import os from 'node:os';
 import path from 'node:path';
 import { execFileSync } from 'node:child_process';
 import { fileURLToPath } from 'node:url';
@@ -52,6 +53,16 @@ function fallbackToMacosSource() {
  process.stdout.write(`Staged macOS helper source fallback: ${macosHelperSourceCopyPath}\n`);
 }

+// Pin the minimum macOS to the app's own floor (Electron's `minos`). Without an
+// explicit target, swiftc stamps the build machine's OS version as the binary's
+// minimum and the helper fails to load on older systems (#213). The arch stays
+// the host's, matching the single-arch app electron-builder packages here.
+const MACOS_HELPER_DEPLOYMENT_TARGET = '12.0';
+
+function macosHelperTarget() {
+  return `${os.arch() === 'x64' ? 'x86_64' : 'arm64'}-apple-macos${MACOS_HELPER_DEPLOYMENT_TARGET}`;
+}
+
 function shouldSkipMacosHelperBuild() {
  return process.env.SUBMINER_SKIP_MACOS_HELPER_BUILD === '1';
 }
@@ -72,9 +83,13 @@ function buildMacosHelper() {
  ensureDir(scriptsOutputDir);

  try {
-    execFileSync('swiftc', ['-O', macosHelperSourcePath, '-o', macosHelperBinaryPath], {
-      stdio: 'inherit',
-    });
+    execFileSync(
+      'swiftc',
+      ['-O', '-target', macosHelperTarget(), macosHelperSourcePath, '-o', macosHelperBinaryPath],
+      {
+        stdio: 'inherit',
+      },
+    );
    fs.chmodSync(macosHelperBinaryPath, 0o755);
    process.stdout.write(`Built macOS helper: ${macosHelperBinaryPath}\n`);
  } catch (error) {
@@ -8,7 +8,7 @@ test('macOS helper build creates dist scripts directory before swiftc output', (
  const buildFunctionIndex = source.indexOf('function buildMacosHelper()');
  assert.notEqual(buildFunctionIndex, -1);

-  const swiftcIndex = source.indexOf("execFileSync('swiftc'", buildFunctionIndex);
+  const swiftcIndex = source.indexOf("'swiftc'", buildFunctionIndex);
  assert.notEqual(swiftcIndex, -1);

  const ensureDirIndex = source.lastIndexOf('ensureDir(scriptsOutputDir)', swiftcIndex);
@@ -18,3 +18,10 @@ test('macOS helper build creates dist scripts directory before swiftc output', (
    'buildMacosHelper must create dist/scripts before swiftc writes the helper binary',
  );
 });
+
+// Regression guard for #213: an untargeted swiftc stamps the build machine's OS
+// version as the helper's minimum, so released builds refuse to load on older macOS.
+test('macOS helper is compiled with an explicit deployment target', () => {
+  assert.match(source, /-target/);
+  assert.match(source, /apple-macos\$\{MACOS_HELPER_DEPLOYMENT_TARGET\}/);
+});
@@ -11,10 +11,12 @@ import type { MediaInput } from './media-input';
 import { AnkiConnectConfig } from './types';

 type TestOverlayNotificationPayload = {
+  id?: string;
  title: string;
  body?: string;
  image?: string;
  variant?: string;
+  persistent?: boolean;
  actions?: Array<{ id: string; label: string; noteId?: number }>;
 };

@@ -153,6 +155,7 @@ function createFieldGroupingMergeCollaborator(options?: {
    getEffectiveSentenceCardConfig: () => ({
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
+      fieldGroupingProvider: 'kiku' as const,
    }),
    getCurrentSubtitleText: () => options?.currentSubtitleText,
    resolveFieldName,
@@ -606,6 +609,7 @@ test('AnkiIntegration applies ready YouTube cache media to every queued note id'
  const integration = new AnkiIntegration(
    {
      fields: {
+        audio: 'ExpressionAudio',
        image: 'Picture',
      },
      media: {
@@ -659,7 +663,7 @@ test('AnkiIntegration applies ready YouTube cache media to every queued note id'
      noteIds.map((noteId) => ({
        noteId,
        fields: {
-          SentenceAudio: { value: '' },
+          ExpressionAudio: { value: '' },
          Picture: { value: '' },
        },
      })),
@@ -944,7 +948,7 @@ test('AnkiIntegration queues YouTube media updates against recovered source URLs
    noteInfo: {
      noteId: 404,
      fields: {
-        SentenceAudio: { value: '' },
+        ExpressionAudio: { value: '' },
        Picture: { value: '' },
      },
    },
@@ -956,7 +960,8 @@ test('AnkiIntegration queues YouTube media updates against recovered source URLs
  assert.equal(queued, true);
  assert.equal(updatedNotes.length, 1);
  assert.equal(updatedNotes[0]?.noteId, 404);
-  assert.match(updatedNotes[0]?.fields.SentenceAudio ?? '', /^\[sound:audio_/);
+  assert.match(updatedNotes[0]?.fields.ExpressionAudio ?? '', /^\[sound:audio_/);
+  assert.equal(updatedNotes[0]?.fields.SentenceAudio, undefined);
  assert.match(updatedNotes[0]?.fields.Picture ?? '', /^<img src="image_/);
  assert.equal(storedMedia.length, 2);
  assert.deepEqual(audioVolumeScales, [0.3 ** 3]);
@@ -1182,6 +1187,117 @@ test('AnkiIntegration embeds generated notification image on overlay mined-card
  assert.deepEqual(cleanupPaths, [notificationIconPath]);
 });

+test('AnkiIntegration keeps overlay card-update progress visible until the terminal notification', async () => {
+  const overlayNotifications: TestOverlayNotificationPayload[] = [];
+  const integration = new AnkiIntegration(
+    {
+      behavior: {
+        notificationType: 'overlay',
+      },
+    },
+    {} as never,
+    {} as never,
+    undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    (payload) => {
+      overlayNotifications.push(payload);
+    },
+  );
+  const updateNotifications = integration as unknown as {
+    beginUpdateProgress: (message: string) => void;
+    showNotification: (noteId: number, label: string | number) => Promise<void>;
+  };
+
+  updateNotifications.beginUpdateProgress('Updating card');
+  await updateNotifications.showNotification(42, '食べる');
+
+  assert.deepEqual(
+    overlayNotifications.map(({ id, variant, persistent }) => ({ id, variant, persistent })),
+    [
+      { id: 'anki-update-progress', variant: 'progress', persistent: true },
+      { id: 'anki-update-progress', variant: 'success', persistent: false },
+    ],
+  );
+});
+
+test('AnkiIntegration dismisses persistent overlay update progress when no terminal notification replaces it', () => {
+  const overlayNotifications: TestOverlayNotificationPayload[] = [];
+  const dismissedIds: string[] = [];
+  const integration = new AnkiIntegration(
+    {
+      behavior: {
+        notificationType: 'overlay',
+      },
+    },
+    {} as never,
+    {} as never,
+    undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    (payload) => {
+      overlayNotifications.push(payload);
+    },
+    undefined,
+    undefined,
+    undefined,
+    (id) => {
+      dismissedIds.push(id);
+    },
+  );
+  const updateNotifications = integration as unknown as {
+    beginUpdateProgress: (message: string) => void;
+    endUpdateProgress: () => void;
+  };
+
+  updateNotifications.beginUpdateProgress('Updating card');
+  updateNotifications.endUpdateProgress();
+
+  assert.equal(overlayNotifications[0]?.persistent, true);
+  assert.deepEqual(dismissedIds, ['anki-update-progress']);
+});
+
+test('AnkiIntegration dismisses overlay update progress after notifications switch to OSD', () => {
+  const behavior: NonNullable<AnkiConnectConfig['behavior']> = {
+    notificationType: 'overlay',
+  };
+  const dismissedIds: string[] = [];
+  const integration = new AnkiIntegration(
+    { behavior },
+    {} as never,
+    {} as never,
+    undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    () => {},
+    undefined,
+    undefined,
+    undefined,
+    (id) => {
+      dismissedIds.push(id);
+    },
+  );
+  const updateNotifications = integration as unknown as {
+    beginUpdateProgress: (message: string) => void;
+    endUpdateProgress: () => void;
+  };
+
+  updateNotifications.beginUpdateProgress('Updating card');
+  behavior.notificationType = 'osd';
+  updateNotifications.endUpdateProgress();
+
+  assert.deepEqual(dismissedIds, ['anki-update-progress']);
+});
+
 test('AnkiIntegration keeps overlay notification image when temp icon write fails', async () => {
  const desktopNotifications: Array<{ title: string; body?: string; icon?: string }> = [];
  const overlayNotifications: TestOverlayNotificationPayload[] = [];
@@ -28,6 +28,8 @@ import {
  KikuMergePreviewResponse,
  NotificationOptions,
  type WordCardKind,
+  type MediaTimingReviewDecision,
+  type MediaTimingReviewRequest,
 } from './types/anki';
 import { AiConfig } from './types/integrations';
 import type { KnownWordMaturityTier } from './types/subtitle';
@@ -218,6 +220,8 @@ export class AnkiIntegration {
    null;
  private overlayNotificationCallback: ((payload: OverlayNotificationPayload) => void) | null =
    null;
+  private overlayNotificationDismissCallback: ((id: string) => void) | null = null;
+  private overlayUpdateProgressActive = false;
  private updateInProgress = false;
  private uiFeedbackState: UiFeedbackState = createUiFeedbackState();
  private parseWarningKeys = new Set<string>();
@@ -238,6 +242,9 @@ export class AnkiIntegration {
  private recordCardsMinedCallback: ((count: number, noteIds?: number[]) => void) | null = null;
  private knownWordCacheUpdatedCallback: (() => void) | null = null;
  private consumeSubtitleMiningContextCallback: (() => SubtitleMiningContext | null) | null = null;
+  private mediaTimingReviewCallback:
+    | ((request: MediaTimingReviewRequest) => Promise<MediaTimingReviewDecision>)
+    | null = null;
  private noteIdRedirects = new Map<number, number>();
  private trackedDuplicateNoteIds = new Map<number, number[]>();
  private getCachedMediaPath: MediaGenerationInputResolverOptions['getCachedMediaPath'] | null =
@@ -265,6 +272,7 @@ export class AnkiIntegration {
    getCachedMediaPath?: MediaGenerationInputResolverOptions['getCachedMediaPath'],
    shouldRequireRemoteMediaCache?: () => boolean,
    getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined,
+    overlayNotificationDismissCallback?: (id: string) => void,
  ) {
    this.config = normalizeAnkiIntegrationConfig(config);
    this.aiConfig = { ...aiConfig };
@@ -280,6 +288,7 @@ export class AnkiIntegration {
    this.getCachedMediaPath = getCachedMediaPath ?? null;
    this.shouldRequireRemoteMediaCache = shouldRequireRemoteMediaCache ?? null;
    this.getYoutubeMediaSourceUrl = getYoutubeMediaSourceUrl ?? null;
+    this.overlayNotificationDismissCallback = overlayNotificationDismissCallback ?? null;
    this.pendingYoutubeMediaQueue = this.createPendingYoutubeMediaQueue();
    this.knownWordCache = this.createKnownWordCache(knownWordCacheStatePath);
    this.pollingRunner = this.createPollingRunner();
@@ -379,8 +388,6 @@ export class AnkiIntegration {
      getCachedMediaPath: this.getCachedMediaPath,
      shouldRequireRemoteMediaCache: () => this.shouldRequireRemoteMediaCache?.() === true,
      getSubtitleMediaRange: (context) => this.getSubtitleMediaRange(context),
-      getResolvedSentenceAudioFieldName: (noteInfo) =>
-        this.getResolvedSentenceAudioFieldName(noteInfo),
      resolveConfiguredFieldName: (noteInfo, ...preferredNames) =>
        this.resolveConfiguredFieldName(noteInfo, ...preferredNames),
      mergeFieldValue: (existing, newValue, overwrite) =>
@@ -509,6 +516,7 @@ export class AnkiIntegration {
        findNotes: async (query, options) =>
          (await this.client.findNotes(query, options)) as number[],
        retrieveMediaFile: (filename) => this.client.retrieveMediaFile(filename),
+        deleteNotes: (noteIds) => this.client.deleteNotes(noteIds),
      },
      mediaGenerator: {
        generateAudio: (
@@ -566,6 +574,7 @@ export class AnkiIntegration {
      getEffectiveSentenceCardConfig: () => this.getEffectiveSentenceCardConfig(),
      getFallbackDurationSeconds: () => this.getFallbackDurationSeconds(),
      appendKnownWordsFromNoteInfo: (noteInfo) => this.appendKnownWordsFromNoteInfo(noteInfo),
+      removeKnownWordNote: (noteId) => this.removeKnownWordNote(noteId),
      isUpdateInProgress: () => this.updateInProgress,
      setUpdateInProgress: (value) => {
        this.updateInProgress = value;
@@ -581,6 +590,7 @@ export class AnkiIntegration {
      recordCardsMinedCallback: (count, noteIds) => {
        this.recordCardsMinedSafely(count, noteIds, 'card creation');
      },
+      reviewMediaTiming: (request) => this.reviewMediaTiming(request),
    });
  }

@@ -637,12 +647,14 @@ export class AnkiIntegration {
        notesInfo: async (noteIds) => (await this.client.notesInfo(noteIds)) as unknown,
        updateNoteFields: (noteId, fields) => this.client.updateNoteFields(noteId, fields),
        storeMediaFile: (filename, data) => this.client.storeMediaFile(filename, data),
+        deleteNotes: (noteIds) => this.client.deleteNotes(noteIds),
      },
      getConfig: () => this.config,
      getCurrentSubtitleText: () => this.mpvClient.currentSubText,
      getCurrentSubtitleStart: () => this.mpvClient.currentSubStart,
      getEffectiveSentenceCardConfig: () => this.getEffectiveSentenceCardConfig(),
      appendKnownWordsFromNoteInfo: (noteInfo) => this.appendKnownWordsFromNoteInfo(noteInfo),
+      removeKnownWordNote: (noteId) => this.removeKnownWordNote(noteId),
      extractFields: (fields) => this.extractFields(fields),
      findDuplicateNote: (expression, excludeNoteId, noteInfo) =>
        this.findDuplicateNote(expression, excludeNoteId, noteInfo),
@@ -657,8 +669,6 @@ export class AnkiIntegration {
        this.setCardTypeFields(updatedFields, availableFieldNames, cardKind),
      resolveConfiguredFieldName: (noteInfo, ...preferredNames) =>
        this.resolveConfiguredFieldName(noteInfo, ...preferredNames),
-      getResolvedSentenceAudioFieldName: (noteInfo) =>
-        this.getResolvedSentenceAudioFieldName(noteInfo),
      getAnimatedImageLeadInSeconds: (noteInfo) => this.getAnimatedImageLeadInSeconds(noteInfo),
      mergeFieldValue: (existing, newValue, overwrite) =>
        this.mergeFieldValue(existing, newValue, overwrite),
@@ -680,6 +690,7 @@ export class AnkiIntegration {
      logWarn: (...args) => log.warn(args[0] as string, ...args.slice(1)),
      logInfo: (...args) => log.info(args[0] as string, ...args.slice(1)),
      logError: (...args) => log.error(args[0] as string, ...args.slice(1)),
+      reviewMediaTiming: (request) => this.reviewMediaTiming(request),
    });
  }

@@ -799,6 +810,12 @@ export class AnkiIntegration {
    }
  }

+  private removeKnownWordNote(noteId: number): void {
+    if (this.knownWordCache.removeNote(noteId)) {
+      this.notifyKnownWordCacheUpdated();
+    }
+  }
+
  private notifyKnownWordCacheUpdated(): void {
    if (!this.knownWordCacheUpdatedCallback) {
      return;
@@ -835,6 +852,19 @@ export class AnkiIntegration {
    };
  }

+  private getSenrenConfig(): {
+    enabled: boolean;
+    fieldGrouping?: 'auto' | 'manual' | 'disabled';
+    deleteDuplicateInAuto?: boolean;
+  } {
+    const senren = this.config.isSenren;
+    return {
+      enabled: senren?.enabled === true,
+      fieldGrouping: senren?.fieldGrouping,
+      deleteDuplicateInAuto: senren?.deleteDuplicateInAuto,
+    };
+  }
+
  private getEffectiveSentenceCardConfig(): {
    model?: string;
    sentenceField: string;
@@ -843,10 +873,27 @@ export class AnkiIntegration {
    kikuEnabled: boolean;
    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
    kikuDeleteDuplicateInAuto: boolean;
+    senrenEnabled: boolean;
+    fieldGroupingProvider: 'kiku' | 'senren' | null;
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
+    fieldGroupingDeleteDuplicateInAuto: boolean;
    wordCardKind: WordCardKind;
  } {
    const lapis = this.getLapisConfig();
    const kiku = this.getKikuConfig();
+    const senren = this.getSenrenConfig();
+
+    const kikuFieldGrouping = (kiku.fieldGrouping || 'disabled') as 'auto' | 'manual' | 'disabled';
+    const senrenFieldGrouping = (senren.fieldGrouping || 'auto') as 'auto' | 'manual' | 'disabled';
+    // Kiku and Senren are mutually exclusive; config resolution enforces it, and
+    // Kiku wins here too in case a runtime patch re-enables both.
+    const fieldGroupingProvider = kiku.enabled ? 'kiku' : senren.enabled ? 'senren' : null;
+    const fieldGroupingMode =
+      fieldGroupingProvider === 'kiku'
+        ? kikuFieldGrouping
+        : fieldGroupingProvider === 'senren'
+          ? senrenFieldGrouping
+          : 'disabled';

    return {
      model: lapis.sentenceCardModel,
@@ -854,8 +901,15 @@ export class AnkiIntegration {
      audioField: 'SentenceAudio',
      lapisEnabled: lapis.enabled,
      kikuEnabled: kiku.enabled,
-      kikuFieldGrouping: (kiku.fieldGrouping || 'disabled') as 'auto' | 'manual' | 'disabled',
+      kikuFieldGrouping,
      kikuDeleteDuplicateInAuto: kiku.deleteDuplicateInAuto !== false,
+      senrenEnabled: senren.enabled,
+      fieldGroupingProvider,
+      fieldGroupingMode,
+      fieldGroupingDeleteDuplicateInAuto:
+        fieldGroupingProvider === 'senren'
+          ? senren.deleteDuplicateInAuto !== false
+          : kiku.deleteDuplicateInAuto !== false,
      wordCardKind: resolveWordCardKindSetting(this.config.lapisKiku?.wordCardKind),
    };
  }
@@ -874,7 +928,7 @@ export class AnkiIntegration {

  private async processNewCard(
    noteId: number,
-    options?: { skipKikuFieldGrouping?: boolean },
+    options?: { skipFieldGrouping?: boolean },
  ): Promise<void> {
    await this.noteUpdateWorkflow.execute(noteId, options);
  }
@@ -1039,7 +1093,7 @@ export class AnkiIntegration {
      videoPath,
      startTime,
      endTime,
-      this.config.media?.audioPadding,
+      context?.mediaPaddingSeconds ?? this.config.media?.audioPadding,
      resolveAudioStreamIndexForMediaGeneration(videoPath, this.mpvClient.currentAudioStreamIndex),
      this.config.media?.normalizeAudio !== false,
      await this.getMpvVolumeScale(),
@@ -1072,7 +1126,7 @@ export class AnkiIntegration {
        videoPath,
        mediaRange.startTime,
        mediaRange.endTime,
-        this.config.media?.audioPadding,
+        context?.mediaPaddingSeconds ?? this.config.media?.audioPadding,
        {
          fps: this.config.media?.animatedFps,
          maxWidth: this.config.media?.animatedMaxWidth,
@@ -1203,12 +1257,13 @@ export class AnkiIntegration {
  private beginUpdateProgress(initialMessage: string): void {
    if (!this.shouldUseOsdNotifications()) {
      if (this.shouldUseOverlayNotifications()) {
+        this.overlayUpdateProgressActive = true;
        this.overlayNotificationCallback?.({
          id: 'anki-update-progress',
          title: 'Anki update',
          body: initialMessage,
          variant: 'progress',
-          persistent: false,
+          persistent: true,
        });
      }
      return;
@@ -1219,6 +1274,10 @@ export class AnkiIntegration {
  }

  private endUpdateProgress(): void {
+    if (this.overlayUpdateProgressActive) {
+      this.overlayUpdateProgressActive = false;
+      this.overlayNotificationDismissCallback?.('anki-update-progress');
+    }
    if (!this.shouldUseOsdNotifications()) {
      return;
    }
@@ -1243,18 +1302,20 @@ export class AnkiIntegration {
    if (!this.shouldUseOsdNotifications()) {
      this.updateInProgress = true;
      if (this.shouldUseOverlayNotifications()) {
+        this.overlayUpdateProgressActive = true;
        this.overlayNotificationCallback?.({
          id: 'anki-update-progress',
          title: 'Anki update',
          body: initialMessage,
          variant: 'progress',
-          persistent: false,
+          persistent: true,
        });
      }
      try {
        return await action();
      } finally {
        this.updateInProgress = false;
+        this.endUpdateProgress();
      }
    }
    return withUpdateProgress(
@@ -1353,6 +1414,7 @@ export class AnkiIntegration {
        : undefined;

    if (shouldShowOverlayNotification && this.overlayNotificationCallback) {
+      this.overlayUpdateProgressActive = false;
      this.overlayNotificationCallback({
        id: 'anki-update-progress',
        title: 'Anki Card Updated',
@@ -1496,7 +1558,7 @@ export class AnkiIntegration {
    trackedDuplicateNoteIdsBeforeCreate: Set<number>,
  ): boolean {
    const sentenceCardConfig = this.getEffectiveSentenceCardConfig();
-    if (!sentenceCardConfig.kikuEnabled || sentenceCardConfig.kikuFieldGrouping === 'disabled') {
+    if (sentenceCardConfig.fieldGroupingMode === 'disabled') {
      return false;
    }

@@ -1555,13 +1617,6 @@ export class AnkiIntegration {
    return sentenceCardConfig.audioField || 'SentenceAudio';
  }

-  private getResolvedSentenceAudioFieldName(noteInfo: NoteInfo): string | null {
-    return (
-      this.resolveNoteFieldName(noteInfo, this.getPreferredSentenceAudioFieldName()) ||
-      this.resolveConfiguredFieldName(noteInfo, this.config.fields?.audio)
-    );
-  }
-
  private getConfiguredWordFieldName(): string {
    return getConfiguredWordFieldName(this.config);
  }
@@ -1723,6 +1778,25 @@ export class AnkiIntegration {
    this.consumeSubtitleMiningContextCallback = callback;
  }

+  setMediaTimingReviewCallback(
+    callback: ((request: MediaTimingReviewRequest) => Promise<MediaTimingReviewDecision>) | null,
+  ): void {
+    this.mediaTimingReviewCallback = callback;
+  }
+
+  private async reviewMediaTiming(
+    request: Omit<MediaTimingReviewRequest, 'audioPadding' | 'maxMediaDuration'>,
+  ): Promise<MediaTimingReviewDecision> {
+    if (this.config.media?.reviewTiming !== true || !this.mediaTimingReviewCallback) {
+      return { action: 'use-original' };
+    }
+    return await this.mediaTimingReviewCallback({
+      ...request,
+      audioPadding: Math.max(0, this.config.media.audioPadding ?? 0),
+      maxMediaDuration: Math.max(0, this.config.media.maxMediaDuration ?? 30),
+    });
+  }
+
  resolveCurrentNoteId(noteId: number): number {
    let resolved = noteId;
    const seen = new Set<number>();
@@ -85,6 +85,7 @@ function createManualUpdateService(overrides: Partial<CardCreationDeps> = {}): {
      },
      findNotes: async () => [42],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => Buffer.from('audio'),
@@ -124,11 +125,11 @@ function createManualUpdateService(overrides: Partial<CardCreationDeps> = {}): {
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -143,7 +144,7 @@ function createManualUpdateService(overrides: Partial<CardCreationDeps> = {}): {
  };
 }

-test('manual clipboard subtitle update replaces sentence audio without touching expression audio', async () => {
+test('manual clipboard subtitle update replaces audio in the configured field', async () => {
  const { service, updatedFields, mergeCalls, storedMedia } = createManualUpdateService();

  await service.updateLastAddedFromClipboard('字幕');
@@ -151,14 +152,144 @@ test('manual clipboard subtitle update replaces sentence audio without touching
  assert.equal(updatedFields.length, 1);
  assert.equal(storedMedia.length, 1);
  const audioValue = `[sound:${storedMedia[0]}]`;
-  assert.equal(updatedFields[0]?.SentenceAudio, audioValue);
-  assert.equal('ExpressionAudio' in updatedFields[0]!, false);
+  assert.equal(updatedFields[0]?.ExpressionAudio, audioValue);
+  assert.equal('SentenceAudio' in updatedFields[0]!, false);
  assert.deepEqual(
    mergeCalls.map((call) => call.overwrite),
    [true],
  );
 });

+test('manual clipboard word-card update uses configured fields with Lapis and Kiku enabled', async () => {
+  const { service, updatedFields } = createManualUpdateService({
+    getConfig: () =>
+      ({
+        deck: 'Mining',
+        fields: {
+          word: 'Expression',
+          sentence: 'Context',
+          audio: 'ContextAudio',
+        },
+        media: {
+          generateAudio: true,
+          generateImage: false,
+          maxMediaDuration: 30,
+        },
+        behavior: {
+          overwriteAudio: false,
+          overwriteImage: false,
+        },
+        ai: false,
+      }) as AnkiConnectConfig,
+    client: {
+      addNote: async () => 0,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: {
+            Expression: { value: '単語' },
+            Sentence: { value: '' },
+            SentenceAudio: { value: '' },
+            Context: { value: '' },
+            ContextAudio: { value: '' },
+          },
+        },
+      ],
+      updateNoteFields: async (_noteId, fields) => {
+        updatedFields.push(fields);
+      },
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [42],
+      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
+    },
+    getEffectiveSentenceCardConfig: () => ({
+      model: 'Sentence',
+      sentenceField: 'Sentence',
+      audioField: 'SentenceAudio',
+      lapisEnabled: true,
+      kikuEnabled: true,
+      fieldGroupingMode: 'disabled',
+    }),
+  });
+
+  await service.updateLastAddedFromClipboard('字幕');
+
+  assert.equal(updatedFields.length, 1);
+  assert.match(updatedFields[0]?.ContextAudio ?? '', /^\[sound:audio_\d+\.mp3\]$/);
+  assert.deepEqual(Object.keys(updatedFields[0] ?? {}).sort(), ['Context', 'ContextAudio']);
+  assert.equal(updatedFields[0]?.Context, '字幕');
+});
+
+test('audio-card action keeps Lapis and Kiku sentence fields', async () => {
+  const { service, updatedFields } = createManualUpdateService({
+    getConfig: () =>
+      ({
+        deck: 'Mining',
+        fields: {
+          word: 'Expression',
+          sentence: 'Context',
+          audio: 'ContextAudio',
+        },
+        media: {
+          generateAudio: true,
+          generateImage: false,
+          maxMediaDuration: 30,
+        },
+        behavior: {},
+        ai: false,
+      }) as AnkiConnectConfig,
+    getMpvClient: () =>
+      ({
+        currentVideoPath: '/video.mp4',
+        currentAudioStreamIndex: 0,
+        currentSubText: '字幕',
+        currentSubStart: 12,
+        currentSubEnd: 14,
+      }) as never,
+    client: {
+      addNote: async () => 0,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: {
+            Expression: { value: '単語' },
+            Sentence: { value: '' },
+            SentenceAudio: { value: '' },
+            Context: { value: '' },
+            ContextAudio: { value: '' },
+          },
+        },
+      ],
+      updateNoteFields: async (_noteId, fields) => {
+        updatedFields.push(fields);
+      },
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [42],
+      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
+    },
+    getEffectiveSentenceCardConfig: () => ({
+      model: 'Sentence',
+      sentenceField: 'Sentence',
+      audioField: 'SentenceAudio',
+      lapisEnabled: true,
+      kikuEnabled: true,
+      fieldGroupingMode: 'disabled',
+    }),
+  });
+
+  await service.markLastCardAsAudioCard();
+
+  assert.equal(updatedFields.length, 1);
+  assert.equal(updatedFields[0]?.Sentence, '字幕');
+  assert.match(updatedFields[0]?.SentenceAudio ?? '', /^\[sound:audio_\d+\.mp3\]$/);
+  assert.equal('Context' in (updatedFields[0] ?? {}), false);
+  assert.equal('ContextAudio' in (updatedFields[0] ?? {}), false);
+});
+
 test('manual clipboard subtitle update marks Kiku word cards as word-and-sentence cards when enabled', async () => {
  const { service, updatedFields } = createManualUpdateService({
    getConfig: () =>
@@ -201,6 +332,7 @@ test('manual clipboard subtitle update marks Kiku word cards as word-and-sentenc
      storeMediaFile: async () => undefined,
      findNotes: async () => [42],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    getEffectiveSentenceCardConfig: () => ({
      model: 'Sentence',
@@ -208,8 +340,7 @@ test('manual clipboard subtitle update marks Kiku word cards as word-and-sentenc
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: true,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    setCardTypeFields,
  });
@@ -225,7 +356,7 @@ test('manual clipboard subtitle update marks Kiku word cards as word-and-sentenc
  });
 });

-test('manual clipboard subtitle update skips audio when sentence audio field is missing', async () => {
+test('manual clipboard subtitle update uses configured audio when SentenceAudio is missing', async () => {
  const { service, updatedFields, mergeCalls, storedMedia } = createManualUpdateService({
    client: {
      addNote: async () => 0,
@@ -248,6 +379,7 @@ test('manual clipboard subtitle update skips audio when sentence audio field is
      },
      findNotes: async () => [42],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
  });

@@ -255,8 +387,9 @@ test('manual clipboard subtitle update skips audio when sentence audio field is

  assert.equal(storedMedia.length, 1);
  assert.equal(updatedFields.length, 1);
-  assert.deepEqual(updatedFields[0], { Sentence: '字幕' });
-  assert.equal(mergeCalls.length, 0);
+  assert.match(updatedFields[0]?.ExpressionAudio ?? '', /^\[sound:audio_\d+\.mp3\]$/);
+  assert.equal(updatedFields[0]?.Sentence, '字幕');
+  assert.equal(mergeCalls.length, 1);
 });

 test('manual clipboard subtitle update uses resolved mpv stream URLs for remote media', async () => {
@@ -335,6 +468,7 @@ test('manual clipboard subtitle update uses resolved mpv stream URLs for remote
      },
      findNotes: async () => [42],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async (path) => {
@@ -383,3 +517,98 @@ test('createSentenceCard relies on Anki progress notification without standalone
  assert.deepEqual(progressMessages, ['Creating sentence card']);
  assert.deepEqual(statusMessages, []);
 });
+
+test('discarding an audio-card timing review deletes the note before evicting its cache entry', async () => {
+  const events: string[] = [];
+  const statusMessages: string[] = [];
+  const { service } = createManualUpdateService({
+    getMpvClient: () =>
+      ({
+        currentVideoPath: '/video.mp4',
+        currentSubText: '字幕',
+        currentSubStart: 4,
+        currentSubEnd: 6,
+        currentTimePos: 5,
+      }) as never,
+    client: {
+      addNote: async () => 0,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: { Expression: { value: '単語' } },
+        },
+      ],
+      updateNoteFields: async () => undefined,
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [42],
+      retrieveMediaFile: async () => '',
+      deleteNotes: async (noteIds) => {
+        events.push(`delete:${noteIds.join(',')}`);
+      },
+    },
+    reviewMediaTiming: async () => ({ action: 'discard' }),
+    removeKnownWordNote: (noteId) => {
+      events.push(`cache:${noteId}`);
+    },
+    showStatusNotification: (message) => {
+      statusMessages.push(message);
+    },
+  });
+
+  await service.markLastCardAsAudioCard();
+
+  assert.deepEqual(events, ['delete:42', 'cache:42']);
+  assert.deepEqual(statusMessages, ['Card deleted.']);
+});
+
+test('keeping an audio card without media skips generation and preserves the note', async () => {
+  let generatedAudio = false;
+  let deleted = false;
+  const updates: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const { service, storedMedia } = createManualUpdateService({
+    getMpvClient: () =>
+      ({
+        currentVideoPath: '/video.mp4',
+        currentSubText: '字幕',
+        currentSubStart: 4,
+        currentSubEnd: 6,
+        currentTimePos: 5,
+      }) as never,
+    client: {
+      addNote: async () => 0,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: { Expression: { value: '単語' }, Sentence: { value: '' } },
+        },
+      ],
+      updateNoteFields: async (noteId, fields) => {
+        updates.push({ noteId, fields });
+      },
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [42],
+      retrieveMediaFile: async () => '',
+      deleteNotes: async () => {
+        deleted = true;
+      },
+    },
+    mediaGenerator: {
+      generateAudio: async () => {
+        generatedAudio = true;
+        return Buffer.from('audio');
+      },
+      generateScreenshot: async () => null,
+      generateAnimatedImage: async () => null,
+    },
+    reviewMediaTiming: async () => ({ action: 'skip-media' }),
+  });
+
+  await service.markLastCardAsAudioCard();
+
+  assert.equal(generatedAudio, false);
+  assert.equal(deleted, false);
+  assert.deepEqual(storedMedia, []);
+  assert.deepEqual(updates, [{ noteId: 42, fields: { Sentence: '字幕' } }]);
+});
@@ -12,6 +12,7 @@ test('sentence card writes generated audio only to sentence audio field', async
  const storedMedia: string[] = [];
  const requestedProperties: string[] = [];
  const audioVolumeScales: Array<number | undefined> = [];
+  const audioRanges: Array<{ start: number; end: number; padding: number | undefined }> = [];

  const deps: CardCreationDeps = {
    getConfig: () =>
@@ -73,17 +74,19 @@ test('sentence card writes generated audio only to sentence audio field', async
      },
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async (
        _path,
-        _startTime,
-        _endTime,
-        _audioPadding,
+        startTime,
+        endTime,
+        audioPadding,
        _audioStreamIndex,
        _normalizeAudio,
        volumeScale,
      ) => {
+        audioRanges.push({ start: startTime, end: endTime, padding: audioPadding });
        audioVolumeScales.push(volumeScale);
        return Buffer.from('audio');
      },
@@ -117,22 +120,19 @@ test('sentence card writes generated audio only to sentence audio field', async
      audioField: 'SentenceAudio',
      lapisEnabled: true,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
+    reviewMediaTiming: async () => ({ action: 'confirm', startTime: 11.4, endTime: 14.2 }),
  };

-  const created = await new CardCreationService(deps).createSentenceCard(
-    '字幕',
-    12,
-    14,
-    'Subtitle',
-  );
+  const service = new CardCreationService(deps);
+  const created = await service.createSentenceCard('字幕', 12, 14, 'Subtitle');

  assert.equal(created, true);
  assert.deepEqual(addedFields[0], {
@@ -144,7 +144,19 @@ test('sentence card writes generated audio only to sentence audio field', async
  assert.equal(storedMedia.length, 1);
  assert.deepEqual(requestedProperties, ['volume']);
  assert.deepEqual(audioVolumeScales, [0.4 ** 3]);
+  assert.deepEqual(audioRanges, [{ start: 11.4, end: 14.2, padding: 0 }]);
  const mediaUpdate = updatedFields.find((fields) => 'SentenceAudio' in fields);
  assert.equal(mediaUpdate?.SentenceAudio, `[sound:${storedMedia[0]}]`);
  assert.equal('ExpressionAudio' in mediaUpdate!, false);
+
+  deps.reviewMediaTiming = async () => ({ action: 'discard' });
+  assert.equal(await service.createSentenceCard('作らない', 20, 22), false);
+  assert.equal(addedFields.length, 1);
+
+  deps.reviewMediaTiming = async () => ({ action: 'skip-media' });
+  assert.equal(await service.createSentenceCard('メディアなし', 30, 32), true);
+  assert.equal(addedFields.length, 2);
+  assert.equal(storedMedia.length, 1);
+  assert.deepEqual(audioRanges, [{ start: 11.4, end: 14.2, padding: 0 }]);
+  assert.deepEqual(requestedProperties, ['volume']);
 });
@@ -42,6 +42,7 @@ test('CardCreationService counts locally created sentence cards', async () => {
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => null,
@@ -69,11 +70,11 @@ test('CardCreationService counts locally created sentence cards', async () => {
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -139,6 +140,7 @@ test('CardCreationService keeps updating after trackLastAddedNoteId throws', asy
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => null,
@@ -168,11 +170,11 @@ test('CardCreationService keeps updating after trackLastAddedNoteId throws', asy
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => {
@@ -238,6 +240,7 @@ test('CardCreationService keeps updating after recordCardsMinedCallback throws',
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => null,
@@ -267,11 +270,11 @@ test('CardCreationService keeps updating after recordCardsMinedCallback throws',
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    recordCardsMinedCallback: () => {
@@ -348,6 +351,7 @@ test('CardCreationService uses stream-open-filename for remote media generation'
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async (path) => {
@@ -387,11 +391,11 @@ test('CardCreationService uses stream-open-filename for remote media generation'
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -454,6 +458,7 @@ test('CardCreationService does not use mpv stream indexes for ready cached YouTu
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async (path, _startTime, _endTime, _padding, audioStreamIndex) => {
@@ -490,11 +495,11 @@ test('CardCreationService does not use mpv stream indexes for ready cached YouTu
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -590,6 +595,7 @@ test('CardCreationService queues YouTube media when required cache is not ready'
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => {
@@ -629,11 +635,11 @@ test('CardCreationService queues YouTube media when required cache is not ready'
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -701,6 +707,7 @@ test('CardCreationService tracks pre-add duplicate note ids for kiku sentence ca
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => null,
@@ -728,11 +735,11 @@ test('CardCreationService tracks pre-add duplicate note ids for kiku sentence ca
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: true,
-      kikuFieldGrouping: 'manual',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'manual',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -790,6 +797,7 @@ test('CardCreationService does not track duplicate ids when pre-add lookup retur
      storeMediaFile: async () => undefined,
      findNotes: async () => [],
      retrieveMediaFile: async () => '',
+      deleteNotes: async () => undefined,
    },
    mediaGenerator: {
      generateAudio: async () => null,
@@ -817,11 +825,11 @@ test('CardCreationService does not track duplicate ids when pre-add lookup retur
      audioField: 'SentenceAudio',
      lapisEnabled: false,
      kikuEnabled: true,
-      kikuFieldGrouping: 'manual',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'manual',
    }),
    getFallbackDurationSeconds: () => 10,
    appendKnownWordsFromNoteInfo: () => undefined,
+    removeKnownWordNote: () => undefined,
    isUpdateInProgress: () => false,
    setUpdateInProgress: () => undefined,
    trackLastAddedNoteId: () => undefined,
@@ -3,7 +3,13 @@ import {
  getConfiguredWordFieldName,
  getPreferredWordValueFromExtractedFields,
 } from '../anki-field-config';
-import { AnkiConnectConfig, type CardKind, type WordCardKind } from '../types/anki';
+import {
+  AnkiConnectConfig,
+  type CardKind,
+  type MediaTimingReviewDecision,
+  type MediaTimingReviewRequest,
+  type WordCardKind,
+} from '../types/anki';
 import { createLogger } from '../logger';
 import type { MediaInput } from '../media-input';
 import { SubtitleTimingTracker } from '../subtitle-timing-tracker';
@@ -55,6 +61,7 @@ interface CardCreationClient {
  storeMediaFile(filename: string, data: Buffer): Promise<void>;
  findNotes(query: string, options?: { maxRetries?: number }): Promise<number[]>;
  retrieveMediaFile(filename: string): Promise<string>;
+  deleteNotes(noteIds: number[]): Promise<void>;
 }

 interface CardCreationMediaGenerator {
@@ -132,18 +139,21 @@ interface CardCreationDeps {
    audioField: string;
    lapisEnabled: boolean;
    kikuEnabled: boolean;
-    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
-    kikuDeleteDuplicateInAuto: boolean;
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
    wordCardKind?: WordCardKind;
  };
  getFallbackDurationSeconds: () => number;
  appendKnownWordsFromNoteInfo: (noteInfo: CardCreationNoteInfo) => void;
+  removeKnownWordNote: (noteId: number) => void;
  isUpdateInProgress: () => boolean;
  setUpdateInProgress: (value: boolean) => void;
  trackLastAddedNoteId?: (noteId: number) => void;
  trackLastAddedDuplicateNoteIds?: (noteId: number, duplicateNoteIds: number[]) => void;
  findDuplicateNoteIds?: (expression: string, noteInfo: CardCreationNoteInfo) => Promise<number[]>;
  recordCardsMinedCallback?: (count: number, noteIds?: number[]) => void;
+  reviewMediaTiming?: (
+    request: Omit<MediaTimingReviewRequest, 'audioPadding' | 'maxMediaDuration'>,
+  ) => Promise<MediaTimingReviewDecision>;
 }

 export class CardCreationService {
@@ -260,9 +270,16 @@ export class CardCreationService {
          fields,
          this.deps.getConfig(),
        );
-        const sentenceAudioField = this.getResolvedSentenceOnlyAudioFieldName(noteInfo);
+        const config = this.deps.getConfig();
+        const sentenceAudioField = this.deps.resolveConfiguredFieldName(
+          noteInfo,
+          config.fields?.audio ?? DEFAULT_ANKI_CONNECT_CONFIG.fields.audio,
+        );
        const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
-        const sentenceField = sentenceCardConfig.sentenceField;
+        const sentenceField = this.deps.resolveConfiguredFieldName(
+          noteInfo,
+          config.fields?.sentence ?? DEFAULT_ANKI_CONNECT_CONFIG.fields.sentence,
+        );

        const sentence = blocks.join(' ');
        const updatedFields: Record<string, string> = {};
@@ -284,7 +301,6 @@ export class CardCreationService {
          `Clipboard update: timing range ${rangeStart.toFixed(2)}s - ${rangeEnd.toFixed(2)}s`,
        );

-        const config = this.deps.getConfig();
        const generateAudio = shouldGenerateAudio(config);
        const generateImage = shouldGenerateImage(config);
        const mediaResolverOptions = this.getMediaResolverOptions();
@@ -451,39 +467,66 @@ export class CardCreationService {
          this.deps.getConfig(),
        );

+        const timingDecision = this.deps.reviewMediaTiming
+          ? await this.deps.reviewMediaTiming({
+              kind: 'audio',
+              text: mpvClient.currentSubText,
+              startTime,
+              endTime,
+              noteId,
+            })
+          : ({ action: 'use-original' } as const);
+        if (timingDecision.action === 'discard') {
+          await this.deps.client.deleteNotes([noteId]);
+          this.deps.removeKnownWordNote(noteId);
+          this.deps.showStatusNotification('Card deleted.');
+          return;
+        }
+        const skipMedia = timingDecision.action === 'skip-media';
+        const exactReviewedRange = timingDecision.action === 'confirm';
+        let sentenceText = mpvClient.currentSubText;
+        if (timingDecision.action === 'confirm') {
+          startTime = timingDecision.startTime;
+          endTime = timingDecision.endTime;
+          sentenceText = timingDecision.text?.trim() || sentenceText;
+        }
+
        const updatedFields: Record<string, string> = {};
        const errors: string[] = [];
        let miscInfoFilename: string | null = null;

        this.deps.setCardTypeFields(updatedFields, Object.keys(noteInfo.fields), 'audio');

-        const sentenceField = this.deps.getConfig().fields?.sentence;
+        const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
+        const sentenceField = sentenceCardConfig.sentenceField;
        if (sentenceField) {
-          const processedSentence = this.deps.processSentence(mpvClient.currentSubText, fields);
+          const processedSentence = this.deps.processSentence(sentenceText, fields);
          updatedFields[sentenceField] = processedSentence;
        }

-        const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
        const audioFieldName = sentenceCardConfig.audioField;
-        try {
-          const audioFilename = this.generateAudioFilename();
-          const audioBuffer = await this.mediaGenerateAudio(
-            mpvClient.currentVideoPath,
-            startTime,
-            endTime,
-          );
+        if (!skipMedia) {
+          try {
+            const audioFilename = this.generateAudioFilename();
+            const audioBuffer = await this.mediaGenerateAudio(
+              mpvClient.currentVideoPath,
+              startTime,
+              endTime,
+              exactReviewedRange ? 0 : undefined,
+            );

-          if (audioBuffer) {
-            await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
-            updatedFields[audioFieldName] = `[sound:${audioFilename}]`;
-            miscInfoFilename = audioFilename;
+            if (audioBuffer) {
+              await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
+              updatedFields[audioFieldName] = `[sound:${audioFilename}]`;
+              miscInfoFilename = audioFilename;
+            }
+          } catch (error) {
+            log.error('Failed to generate audio for audio card:', (error as Error).message);
+            errors.push('audio');
          }
-        } catch (error) {
-          log.error('Failed to generate audio for audio card:', (error as Error).message);
-          errors.push('audio');
        }

-        if (shouldGenerateImage(this.deps.getConfig())) {
+        if (!skipMedia && shouldGenerateImage(this.deps.getConfig())) {
          try {
            const animatedLeadInSeconds = await this.deps.getAnimatedImageLeadInSeconds(noteInfo);
            const imageFilename = this.generateImageFilename();
@@ -492,6 +535,7 @@ export class CardCreationService {
              startTime,
              endTime,
              animatedLeadInSeconds,
+              exactReviewedRange,
            );

            const imageField = this.deps.getConfig().fields?.image;
@@ -564,9 +608,29 @@ export class CardCreationService {

    try {
      return await this.deps.withUpdateProgress('Creating sentence card', async () => {
+        const timingDecision = this.deps.reviewMediaTiming
+          ? await this.deps.reviewMediaTiming({
+              kind: 'sentence',
+              text: sentence,
+              startTime,
+              endTime,
+            })
+          : ({ action: 'use-original' } as const);
+        if (timingDecision.action === 'discard') {
+          this.deps.showStatusNotification('Card creation cancelled.');
+          return false;
+        }
+        const skipMedia = timingDecision.action === 'skip-media';
+        const exactReviewedRange = timingDecision.action === 'confirm';
+        if (timingDecision.action === 'confirm') {
+          startTime = timingDecision.startTime;
+          endTime = timingDecision.endTime;
+          sentence = timingDecision.text?.trim() || sentence;
+        }
+
        const config = this.deps.getConfig();
-        const generateAudio = shouldGenerateAudio(config);
-        const generateImage = shouldGenerateImage(config);
+        const generateAudio = !skipMedia && shouldGenerateAudio(config);
+        const generateImage = !skipMedia && shouldGenerateImage(config);
        const mediaResolverOptions = this.getMediaResolverOptions();
        const videoPath = generateImage
          ? await resolveMediaGenerationInput(mpvClient, 'video', mediaResolverOptions)
@@ -632,8 +696,7 @@ export class CardCreationService {
        ).trim();
        let duplicateNoteIds: number[] = [];
        if (
-          sentenceCardConfig.kikuEnabled &&
-          sentenceCardConfig.kikuFieldGrouping !== 'disabled' &&
+          sentenceCardConfig.fieldGroupingMode !== 'disabled' &&
          pendingExpressionText &&
          this.deps.findDuplicateNoteIds
        ) {
@@ -732,6 +795,7 @@ export class CardCreationService {
            generateAudio,
            generateImage,
            volumeScale,
+            ...(exactReviewedRange ? { mediaPaddingSeconds: 0 } : {}),
          });
          await this.deps.showNotification(noteId, label, 'media queued');
          return true;
@@ -747,7 +811,12 @@ export class CardCreationService {
          try {
            const audioFilename = this.generateAudioFilename();
            const audioBuffer = audioSourcePath
-              ? await this.mediaGenerateAudio(audioSourcePath, startTime, endTime)
+              ? await this.mediaGenerateAudio(
+                  audioSourcePath,
+                  startTime,
+                  endTime,
+                  exactReviewedRange ? 0 : undefined,
+                )
              : null;

            if (audioBuffer) {
@@ -765,7 +834,13 @@ export class CardCreationService {
        if (generateImage) {
          try {
            const imageFilename = this.generateImageFilename();
-            const imageBuffer = await this.generateImageBuffer(videoPath!, startTime, endTime);
+            const imageBuffer = await this.generateImageBuffer(
+              videoPath!,
+              startTime,
+              endTime,
+              0,
+              exactReviewedRange,
+            );

            const imageField = config.fields?.image;
            if (imageBuffer && imageField) {
@@ -806,22 +881,6 @@ export class CardCreationService {
    }
  }

-  private getResolvedSentenceAudioFieldName(noteInfo: CardCreationNoteInfo): string | null {
-    return (
-      this.deps.resolveNoteFieldName(
-        noteInfo,
-        this.deps.getEffectiveSentenceCardConfig().audioField || 'SentenceAudio',
-      ) || this.deps.resolveConfiguredFieldName(noteInfo, this.deps.getConfig().fields?.audio)
-    );
-  }
-
-  private getResolvedSentenceOnlyAudioFieldName(noteInfo: CardCreationNoteInfo): string | null {
-    return this.deps.resolveNoteFieldName(
-      noteInfo,
-      this.deps.getEffectiveSentenceCardConfig().audioField || 'SentenceAudio',
-    );
-  }
-
  private createPendingNoteInfo(fields: Record<string, string>): CardCreationNoteInfo {
    return {
      noteId: -1,
@@ -833,6 +892,7 @@ export class CardCreationService {
    videoPath: MediaInput,
    startTime: number,
    endTime: number,
+    audioPaddingOverride?: number,
  ): Promise<Buffer | null> {
    const mpvClient = this.deps.getMpvClient();
    if (!mpvClient) {
@@ -843,7 +903,7 @@ export class CardCreationService {
      videoPath,
      startTime,
      endTime,
-      this.deps.getConfig().media?.audioPadding,
+      audioPaddingOverride ?? this.deps.getConfig().media?.audioPadding,
      resolveAudioStreamIndexForMediaGeneration(
        videoPath,
        mpvClient.currentAudioStreamIndex ?? undefined,
@@ -861,13 +921,16 @@ export class CardCreationService {
    startTime: number,
    endTime: number,
    animatedLeadInSeconds = 0,
+    exactReviewedRange = false,
  ): Promise<Buffer | null> {
    const mpvClient = this.deps.getMpvClient();
    if (!mpvClient) {
      return null;
    }

-    const timestamp = mpvClient.currentTimePos || 0;
+    const timestamp = exactReviewedRange
+      ? startTime + (endTime - startTime) / 2
+      : mpvClient.currentTimePos || 0;

    if (this.deps.getConfig().media?.imageType === 'avif') {
      let imageStart = startTime;
@@ -883,7 +946,7 @@ export class CardCreationService {
        videoPath,
        imageStart,
        imageEnd,
-        this.deps.getConfig().media?.audioPadding,
+        exactReviewedRange ? 0 : this.deps.getConfig().media?.audioPadding,
        {
          fps: this.deps.getConfig().media?.animatedFps,
          maxWidth: this.deps.getConfig().media?.animatedMaxWidth,
@@ -26,6 +26,7 @@ function createCollaborator(
      miscInfoValue?: string;
    };
    warnings?: Array<{ fieldName: string; reason: string; detail?: string }>;
+    fieldGroupingProvider?: 'kiku' | 'senren' | null;
  } = {},
 ) {
  const warnings = options.warnings ?? [];
@@ -46,6 +47,8 @@ function createCollaborator(
      getEffectiveSentenceCardConfig: () => ({
        sentenceField: 'Sentence',
        audioField: 'SentenceAudio',
+        fieldGroupingProvider:
+          options.fieldGroupingProvider === undefined ? 'kiku' : options.fieldGroupingProvider,
      }),
      getCurrentSubtitleText: () => options.currentSubtitleText,
      resolveFieldName,
@@ -251,7 +254,218 @@ test('computeFieldGroupingMergedFields uses generated media only when includeGen
  assert.equal(withMedia.MiscInfo, '<span data-group-id="11">generated misc</span>');
 });

-test('computeFieldGroupingMergedFields clears SentenceFurigana when either note lacks it', async () => {
+test('computeFieldGroupingMergedFields merges Senren notes into scene-switching markup', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      word: '語',
+      sentence: '<span class="group">前<span class="highlight">語</span>後</span>',
+      sentenceAudio: '[sound:original.opus]',
+      picture: '<img src="original.webp">',
+      miscInfo: '<span class="group">Show EP1 (0:01:00)</span>',
+    }),
+    makeNote(200, {
+      word: '語',
+      sentence: '<span class="group">次<span class="highlight">語</span>文</span>',
+      sentenceAudio: '[sound:new.opus]',
+      picture: '<img src="new.webp">',
+      miscInfo: 'Show EP2 (0:02:00)',
+    }),
+    false,
+  );
+
+  assert.equal(
+    merged.sentence,
+    '<span class="group">前<span class="highlight">語</span>後</span>' +
+      '<span class="group2">次<span class="highlight">語</span>文</span>',
+  );
+  assert.equal(merged.sentenceAudio, '[sound:original.opus][sound:new.opus]');
+  assert.equal(merged.picture, '<img src="original.webp"><img src="new.webp">');
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">Show EP1 (0:01:00)</span><span class="group2">Show EP2 (0:02:00)</span>',
+  );
+});
+
+test('Senren merge warns for invalid source audio when kept audio is empty', async () => {
+  const warnings: Array<{ fieldName: string; reason: string; detail?: string }> = [];
+  const { collaborator } = createCollaborator({
+    fieldGroupingProvider: 'senren',
+    warnings,
+  });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, { SentenceAudio: '' }),
+    makeNote(200, { SentenceAudio: 'invalid audio' }),
+    false,
+  );
+
+  assert.equal(merged.SentenceAudio, 'invalid audio');
+  assert.deepEqual(warnings, [
+    {
+      fieldName: 'SentenceAudio',
+      reason: 'missing-sound-tag',
+      detail: undefined,
+    },
+  ]);
+});
+
+test('Senren merge wraps ungrouped legacy content and preserves numbered groups', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      sentence: 'plain legacy sentence',
+      sentenceAudio: '[sound:a.opus][sound:b.opus]',
+      miscInfo: '<span class="group2">pinned</span> stray text',
+    }),
+    makeNote(200, {
+      sentence: '<span class="group">new sentence</span>',
+      sentenceAudio: '[sound:c.opus]',
+      miscInfo: '<span class="group">new misc</span>',
+    }),
+    false,
+  );
+
+  assert.equal(
+    merged.sentence,
+    '<span class="group">plain legacy sentence</span><span class="group3">new sentence</span>',
+  );
+  assert.equal(merged.sentenceAudio, '[sound:a.opus][sound:b.opus][sound:c.opus]');
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group2">pinned</span><span class="group">stray text</span>' +
+      '<span class="group3">new misc</span>',
+  );
+});
+
+test('Senren merge rebases numbered groups from an appended source note', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      sentenceAudio: '[sound:keep-a.opus][sound:keep-b.opus]',
+      miscInfo: '<span class="group">keep one</span><span class="group">keep two</span>',
+    }),
+    makeNote(200, {
+      sentenceAudio: '[sound:source-a.opus][sound:source-b.opus]',
+      miscInfo: '<span class="group2">source two</span>',
+    }),
+    false,
+  );
+
+  assert.equal(
+    merged.sentenceAudio,
+    '[sound:keep-a.opus][sound:keep-b.opus][sound:source-a.opus][sound:source-b.opus]',
+  );
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">keep one</span><span class="group">keep two</span>' +
+      '<span class="group4">source two</span>',
+  );
+});
+
+test('Senren merge rebases plain source groups after empty and sparse kept fields', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      sentenceAudio: '[sound:keep-a.opus][sound:keep-b.opus]',
+      sentence: '',
+      miscInfo: '<span class="group">keep first</span>',
+    }),
+    makeNote(200, {
+      sentenceAudio: '[sound:source-a.opus][sound:source-b.opus]',
+      sentence: '<span class="group">source first</span>',
+      miscInfo: '<span class="group">source first</span><span class="group2">source second</span>',
+    }),
+    false,
+  );
+
+  assert.equal(merged.sentence, '<span class="group3">source first</span>');
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">keep first</span><span class="group3">source first</span>' +
+      '<span class="group4">source second</span>',
+  );
+});
+
+test('Senren merge keeps ungrouped text in place around an existing group span', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      miscInfo: 'leading<span class="group">middle</span>trailing',
+      sentenceAudio: '[sound:a.opus][sound:b.opus][sound:c.opus]',
+    }),
+    makeNote(200, {
+      miscInfo: '<span class="group">appended</span>',
+      sentenceAudio: '[sound:d.opus]',
+    }),
+    false,
+  );
+
+  // Order must follow the source field, and the two ungrouped runs must stay separate.
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">leading</span><span class="group">middle</span>' +
+      '<span class="group">trailing</span><span class="group4">appended</span>',
+  );
+  assert.equal(merged.sentenceAudio, '[sound:a.opus][sound:b.opus][sound:c.opus][sound:d.opus]');
+});
+
+test('Senren merge closes unclosed group spans so later scenes stay siblings', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, { miscInfo: '<span class="group">a<span class="highlight">b' }),
+    makeNote(200, { miscInfo: '<span class="group">next</span>' }),
+    false,
+  );
+
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">a<span class="highlight">b</span></span><span class="group">next</span>',
+  );
+  const openTags = merged.miscInfo!.match(/<span\b/g)?.length ?? 0;
+  const closeTags = merged.miscInfo!.match(/<\/span>/g)?.length ?? 0;
+  assert.equal(openTags, closeTags);
+});
+
+test('Senren merge closes unclosed trailing markup before appending later scenes', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, { miscInfo: 'leading<span class="highlight">tail' }),
+    makeNote(200, { miscInfo: '<span class="group">next</span>' }),
+    false,
+  );
+
+  assert.equal(
+    merged.miscInfo,
+    '<span class="group">leading<span class="highlight">tail</span></span>' +
+      '<span class="group">next</span>',
+  );
+});
+
+test('Kiku merge clears SentenceFurigana when either note lacks it', async () => {
  const { collaborator } = createCollaborator();

  const merged = await collaborator.computeFieldGroupingMergedFields(
@@ -268,3 +482,21 @@ test('computeFieldGroupingMergedFields clears SentenceFurigana when either note

  assert.equal(merged.SentenceFurigana, '');
 });
+
+test('Senren merge keeps duplicate SentenceFurigana when the kept field is empty', async () => {
+  const { collaborator } = createCollaborator({ fieldGroupingProvider: 'senren' });
+
+  const merged = await collaborator.computeFieldGroupingMergedFields(
+    300,
+    200,
+    makeNote(300, {
+      SentenceFurigana: '',
+    }),
+    makeNote(200, {
+      SentenceFurigana: 'duplicate furigana',
+    }),
+    false,
+  );
+
+  assert.equal(merged.SentenceFurigana, '<span class="group">duplicate furigana</span>');
+});
@@ -19,6 +19,7 @@ interface FieldGroupingMergeDeps {
  getEffectiveSentenceCardConfig: () => {
    sentenceField: string;
    audioField: string;
+    fieldGroupingProvider: 'kiku' | 'senren' | null;
  };
  getCurrentSubtitleText: () => string | undefined;
  resolveFieldName: (availableFieldNames: string[], preferredName: string) => string | null;
@@ -78,6 +79,13 @@ export class FieldGroupingMergeCollaborator {
    const configuredWordField = getConfiguredWordFieldName(config);
    const groupableFields = this.getGroupableFieldNames();
    const keepFieldNames = Object.keys(keepNoteInfo.fields);
+    const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
+    const senrenSourceSceneOffset =
+      sentenceCardConfig.fieldGroupingProvider === 'senren'
+        ? this.countSenrenAudioScenes(
+            this.getResolvedFieldValue(keepNoteInfo, sentenceCardConfig.audioField),
+          )
+        : 0;
    const sourceFields: Record<string, string> = {};
    const resolvedKeepFieldByPreferred = new Map<string, string>();
    for (const preferredFieldName of groupableFields) {
@@ -154,14 +162,18 @@ export class FieldGroupingMergeCollaborator {
      if (!existingValue.trim() && !newValue.trim()) continue;

      if (keepFieldNormalized === 'sentencefurigana') {
+        const hasBothValues = existingValue.trim().length > 0 && newValue.trim().length > 0;
+        const usesSenrenGrouping =
+          this.deps.getEffectiveSentenceCardConfig().fieldGroupingProvider === 'senren';
        mergedFields[keepFieldName] =
-          existingValue.trim() && newValue.trim()
+          hasBothValues || usesSenrenGrouping
            ? this.applyFieldGrouping(
                existingValue,
                newValue,
                keepNoteId,
                deleteNoteId,
                keepFieldName,
+                senrenSourceSceneOffset,
              )
            : '';
        continue;
@@ -174,6 +186,7 @@ export class FieldGroupingMergeCollaborator {
          keepNoteId,
          deleteNoteId,
          keepFieldName,
+          senrenSourceSceneOffset,
        );
      } else if (existingValue.trim() && newValue.trim()) {
        mergedFields[keepFieldName] = this.applyFieldGrouping(
@@ -182,6 +195,7 @@ export class FieldGroupingMergeCollaborator {
          keepNoteId,
          deleteNoteId,
          keepFieldName,
+          senrenSourceSceneOffset,
        );
      } else {
        if (!newValue.trim()) continue;
@@ -342,13 +356,152 @@ export class FieldGroupingMergeCollaborator {
    return [...entries].sort((a, b) => b.groupId - a.groupId);
  }

+  private isSentenceAudioField(fieldName: string): boolean {
+    const normalized = fieldName.toLowerCase();
+    const audioField = (
+      this.deps.getEffectiveSentenceCardConfig().audioField || 'sentenceaudio'
+    ).toLowerCase();
+    return normalized === 'sentenceaudio' || normalized === audioField;
+  }
+
+  private isSenrenGroupOpenTag(openTag: string): boolean {
+    const classMatch =
+      openTag.match(/class\s*=\s*"([^"]*)"/i) || openTag.match(/class\s*=\s*'([^']*)'/i);
+    if (!classMatch) return false;
+    // Senren's templates match class tokens case-sensitively (/^group\d*$/).
+    return classMatch[1]!.split(/\s+/).some((token) => /^group\d*$/.test(token));
+  }
+
+  private countSenrenAudioScenes(value: string): number {
+    const soundEntries = value.match(/\[sound:[^\]]+\]/g)?.length ?? 0;
+    if (soundEntries > 0) return soundEntries;
+    return this.parseSenrenSceneEntries(value).length;
+  }
+
+  private rebaseSenrenGroup(entry: string, sceneOffset: number, sourceEntryIndex: number): string {
+    if (sceneOffset <= 0) return entry;
+
+    return entry.replace(
+      /^(\s*<span\b[^>]*?\bclass\s*=\s*)(["'])([^"']*)\2/i,
+      (_match: string, prefix: string, quote: string, rawClasses: string) => {
+        const classes = rawClasses
+          .split(/(\s+)/)
+          .map((classToken) => {
+            if (classToken === 'group') {
+              return `group${sceneOffset + sourceEntryIndex + 1}`;
+            }
+            const groupMatch = classToken.match(/^group(\d+)$/);
+            if (!groupMatch) return classToken;
+            const targetScene = Number(groupMatch[1]);
+            if (!Number.isSafeInteger(targetScene) || targetScene <= 0) return classToken;
+            return `group${targetScene + sceneOffset}`;
+          })
+          .join('');
+        return `${prefix}${quote}${classes}${quote}`;
+      },
+    );
+  }
+
+  /**
+   * Splits a Senren field into ordered scene entries. Top-level
+   * `<span class="group">`/`"groupN"` spans are kept verbatim (nested markup like
+   * `<span class="highlight">` included); ungrouped runs are wrapped in a group
+   * span at their original position, because Senren discards anything outside a
+   * group span once scene switching activates.
+   */
+  private parseSenrenSceneEntries(value: string): string[] {
+    const tokenRegex = /<span\b[^>]*>|<\/span>/gi;
+    const entries: string[] = [];
+    const pushUngrouped = (raw: string): void => {
+      const text = raw.replace(/<br\s*\/?>/gi, ' ').trim();
+      if (text) entries.push(`<span class="group">${text}</span>`);
+    };
+    let cursor = 0;
+    let depth = 0;
+    let entryStart = -1;
+    let match;
+    while ((match = tokenRegex.exec(value)) !== null) {
+      const token = match[0]!;
+      if (token[1] !== '/') {
+        if (depth === 0 && this.isSenrenGroupOpenTag(token)) {
+          pushUngrouped(value.slice(cursor, match.index));
+          entryStart = match.index;
+          cursor = match.index;
+        }
+        depth += 1;
+      } else {
+        depth = Math.max(0, depth - 1);
+        if (depth === 0 && entryStart !== -1) {
+          const end = match.index + token.length;
+          entries.push(value.slice(entryStart, end));
+          entryStart = -1;
+          cursor = end;
+        }
+      }
+    }
+    if (entryStart !== -1) {
+      // Unclosed group span: close every span still open (the group and any nested
+      // markup) so the following scenes are siblings rather than nested inside it.
+      entries.push(`${value.slice(entryStart)}${'</span>'.repeat(depth)}`);
+    } else {
+      pushUngrouped(`${value.slice(cursor)}${'</span>'.repeat(depth)}`);
+    }
+    return entries;
+  }
+
+  /**
+   * Merges two notes' field values in Senren's scene-switching format. Scenes are
+   * appended in order (existing first, never resorted) so indices stay aligned
+   * across sentence/picture/miscInfo with the sentenceAudio entries, which alone
+   * drive Senren's scene count.
+   */
+  private applySenrenFieldGrouping(
+    existingValue: string,
+    newValue: string,
+    fieldName: string,
+    sourceSceneOffset: number,
+  ): string {
+    if (this.isPictureField(fieldName)) {
+      const tags = [...this.extractImageTags(existingValue), ...this.extractImageTags(newValue)];
+      if (tags.length === 0) return existingValue || newValue;
+      return tags.join('');
+    }
+
+    if (this.isSentenceAudioField(fieldName)) {
+      const existing = existingValue.trim();
+      const added = newValue.trim();
+      if (added && !/\[sound:[^\]]+\]/.test(added)) {
+        this.deps.warnFieldParseOnce(fieldName, 'missing-sound-tag');
+      }
+      if (!existing || !added) return existing || added;
+      return existing + added;
+    }
+
+    const sourceEntries = this.parseSenrenSceneEntries(newValue).map((entry, sourceEntryIndex) =>
+      this.rebaseSenrenGroup(entry, sourceSceneOffset, sourceEntryIndex),
+    );
+    const merged = [...this.parseSenrenSceneEntries(existingValue), ...sourceEntries];
+    if (merged.length === 0) return existingValue || newValue;
+    return merged.join('');
+  }
+
  private applyFieldGrouping(
    existingValue: string,
    newValue: string,
    keepGroupId: number,
    sourceGroupId: number,
    fieldName: string,
+    senrenSourceSceneOffset: number,
  ): string {
+    if (this.deps.getEffectiveSentenceCardConfig().fieldGroupingProvider === 'senren') {
+      return this.applySenrenFieldGrouping(
+        existingValue,
+        newValue,
+        fieldName,
+        senrenSourceSceneOffset,
+      );
+    }
+
    if (this.shouldUseStrictSpanGrouping(fieldName)) {
      if (this.isPictureField(fieldName)) {
        const keepEntries = this.parsePictureEntries(existingValue, keepGroupId);
@@ -71,7 +71,7 @@ function createWorkflowHarness() {
    getEffectiveSentenceCardConfig: () => ({
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingDeleteDuplicateInAuto: true,
    }),
    getCurrentSubtitleText: () => 'subtitle-text',
    getFieldGroupingCallback: (): FieldGroupingCallback | null => {
@@ -24,7 +24,7 @@ export interface FieldGroupingWorkflowDeps {
  getEffectiveSentenceCardConfig: () => {
    sentenceField: string;
    audioField: string;
-    kikuDeleteDuplicateInAuto: boolean;
+    fieldGroupingDeleteDuplicateInAuto: boolean;
  };
  getCurrentSubtitleText: () => string | undefined;
  getFieldGroupingCallback:
@@ -75,7 +75,7 @@ export class FieldGroupingWorkflow {
        originalNoteId,
        newNoteId,
        this.getExpression(newNoteInfo),
-        sentenceCardConfig.kikuDeleteDuplicateInAuto,
+        sentenceCardConfig.fieldGroupingDeleteDuplicateInAuto,
      );
    } catch (error) {
      this.deps.logError('Field grouping auto merge failed:', (error as Error).message);
@@ -21,14 +21,14 @@ function createHarness(
    manualHandled?: boolean;
    expression?: string | null;
    currentSentenceImageField?: string | undefined;
-    onProcessNewCard?: (noteId: number, options?: { skipKikuFieldGrouping?: boolean }) => void;
+    onProcessNewCard?: (noteId: number, options?: { skipFieldGrouping?: boolean }) => void;
  } = {},
 ) {
  const calls: string[] = [];
  const findNotesQueries: Array<{ query: string; maxRetries?: number }> = [];
  const noteInfoRequests: number[][] = [];
  const duplicateRequests: Array<{ expression: string; excludeNoteId: number }> = [];
-  const processCalls: Array<{ noteId: number; options?: { skipKikuFieldGrouping?: boolean } }> = [];
+  const processCalls: Array<{ noteId: number; options?: { skipFieldGrouping?: boolean } }> = [];
  const autoCalls: Array<{ originalNoteId: number; newNoteId: number; expression: string }> = [];
  const manualCalls: Array<{ originalNoteId: number; newNoteId: number; expression: string }> = [];

@@ -46,9 +46,8 @@ function createHarness(
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
      lapisEnabled: false,
-      kikuEnabled: options.kikuEnabled ?? true,
-      kikuFieldGrouping: options.kikuFieldGrouping ?? 'auto',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingProvider: (options.kikuEnabled ?? true) ? ('kiku' as const) : null,
+      fieldGroupingMode: options.kikuFieldGrouping ?? 'auto',
    }),
    isUpdateInProgress: () => false,
    getDeck: options.deck ? () => options.deck : undefined,
@@ -134,7 +133,7 @@ test('triggerFieldGroupingForLastAddedCard stops when kiku mode is disabled', as

  await harness.service.triggerFieldGroupingForLastAddedCard();

-  assert.deepEqual(harness.calls, ['osd:Kiku mode is not enabled']);
+  assert.deepEqual(harness.calls, ['osd:Field grouping requires Kiku or Senren mode']);
  assert.equal(harness.findNotesQueries.length, 0);
 });

@@ -143,7 +142,7 @@ test('triggerFieldGroupingForLastAddedCard stops when field grouping is disabled

  await harness.service.triggerFieldGroupingForLastAddedCard();

-  assert.deepEqual(harness.calls, ['osd:Kiku field grouping is disabled']);
+  assert.deepEqual(harness.calls, ['osd:Field grouping is disabled']);
  assert.equal(harness.findNotesQueries.length, 0);
 });

@@ -155,9 +154,8 @@ test('triggerFieldGroupingForLastAddedCard stops when an update is already in pr
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
      lapisEnabled: false,
-      kikuEnabled: true,
-      kikuFieldGrouping: 'auto',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingProvider: 'kiku' as const,
+      fieldGroupingMode: 'auto' as const,
    }),
    isUpdateInProgress: () => true,
    withUpdateProgress: async () => {
@@ -266,7 +264,7 @@ test('triggerFieldGroupingForLastAddedCard prefers tracked duplicate note ids be
 });

 test('triggerFieldGroupingForLastAddedCard refreshes the card when configured fields are missing', async () => {
-  const processCalls: Array<{ noteId: number; options?: { skipKikuFieldGrouping?: boolean } }> = [];
+  const processCalls: Array<{ noteId: number; options?: { skipFieldGrouping?: boolean } }> = [];
  const harness = createHarness({
    noteIds: [11],
    notesInfo: [
@@ -298,7 +296,7 @@ test('triggerFieldGroupingForLastAddedCard refreshes the card when configured fi

  await harness.service.triggerFieldGroupingForLastAddedCard();

-  assert.deepEqual(processCalls, [{ noteId: 11, options: { skipKikuFieldGrouping: true } }]);
+  assert.deepEqual(processCalls, [{ noteId: 11, options: { skipFieldGrouping: true } }]);
  assert.deepEqual(harness.manualCalls, []);
 });

@@ -352,9 +350,8 @@ test('buildFieldGroupingPreview returns merged compact and full previews', async
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
      lapisEnabled: false,
-      kikuEnabled: true,
-      kikuFieldGrouping: 'auto',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingProvider: 'kiku' as const,
+      fieldGroupingMode: 'auto' as const,
    }),
    isUpdateInProgress: () => false,
    withUpdateProgress: async (_message, action) => action(),
@@ -417,9 +414,8 @@ test('buildFieldGroupingPreview reports missing notes cleanly', async () => {
      sentenceField: 'Sentence',
      audioField: 'SentenceAudio',
      lapisEnabled: false,
-      kikuEnabled: true,
-      kikuFieldGrouping: 'auto',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingProvider: 'kiku' as const,
+      fieldGroupingMode: 'auto' as const,
    }),
    isUpdateInProgress: () => false,
    withUpdateProgress: async (_message, action) => action(),
@@ -20,9 +20,8 @@ interface FieldGroupingDeps {
    sentenceField: string;
    audioField: string;
    lapisEnabled: boolean;
-    kikuEnabled: boolean;
-    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
-    kikuDeleteDuplicateInAuto: boolean;
+    fieldGroupingProvider: 'kiku' | 'senren' | null;
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
  };
  isUpdateInProgress: () => boolean;
  getDeck?: () => string | undefined;
@@ -46,7 +45,7 @@ interface FieldGroupingDeps {
    noteInfo: FieldGroupingNoteInfo,
    configuredFieldNames: (string | undefined)[],
  ) => boolean;
-  processNewCard: (noteId: number, options?: { skipKikuFieldGrouping?: boolean }) => Promise<void>;
+  processNewCard: (noteId: number, options?: { skipFieldGrouping?: boolean }) => Promise<void>;
  getSentenceCardImageFieldName: () => string | undefined;
  resolveFieldName: (availableFieldNames: string[], preferredName: string) => string | null;
  computeFieldGroupingMergedFields: (
@@ -76,12 +75,12 @@ export class FieldGroupingService {

  async triggerFieldGroupingForLastAddedCard(): Promise<void> {
    const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
-    if (!sentenceCardConfig.kikuEnabled) {
-      this.deps.showOsdNotification('Kiku mode is not enabled');
+    if (sentenceCardConfig.fieldGroupingProvider === null) {
+      this.deps.showOsdNotification('Field grouping requires Kiku or Senren mode');
      return;
    }
-    if (sentenceCardConfig.kikuFieldGrouping === 'disabled') {
-      this.deps.showOsdNotification('Kiku field grouping is disabled');
+    if (sentenceCardConfig.fieldGroupingMode === 'disabled') {
+      this.deps.showOsdNotification('Field grouping is disabled');
      return;
    }

@@ -134,7 +133,7 @@ export class FieldGroupingService {
          ])
        ) {
          await this.deps.processNewCard(noteId, {
-            skipKikuFieldGrouping: true,
+            skipFieldGrouping: true,
          });
        }

@@ -147,7 +146,7 @@ export class FieldGroupingService {

        const noteInfo = refreshedInfo[0]!;

-        if (sentenceCardConfig.kikuFieldGrouping === 'auto') {
+        if (sentenceCardConfig.fieldGroupingMode === 'auto') {
          await this.deps.handleFieldGroupingAuto(
            duplicateNoteId,
            noteId,
@@ -261,6 +261,32 @@ test('KnownWordCacheManager invalidates persisted cache when fields.word changes
  }
 });

+test('KnownWordCacheManager removes a deleted note from memory and persisted state', () => {
+  const config: AnkiConnectConfig = {
+    fields: { word: 'Word' },
+    knownWords: { highlightEnabled: true },
+  };
+  const { manager, statePath, cleanup } = createKnownWordCacheHarness(config);
+
+  try {
+    manager.appendFromNoteInfo({
+      noteId: 42,
+      fields: { Word: { value: '猫' } },
+    });
+
+    assert.equal(manager.removeNote(42), true);
+    assert.equal(manager.removeNote(42), false);
+    assert.equal(manager.isKnownWord('猫'), false);
+
+    const persisted = JSON.parse(fs.readFileSync(statePath, 'utf-8')) as {
+      notes?: Record<string, unknown>;
+    };
+    assert.deepEqual(persisted.notes, {});
+  } finally {
+    cleanup();
+  }
+});
+
 test('KnownWordCacheManager refresh incrementally reconciles deleted and edited note words', async () => {
  const config: AnkiConnectConfig = {
    fields: {
@@ -350,6 +350,17 @@ export class KnownWordCacheManager {
    return true;
  }

+  removeNote(noteId: number): boolean {
+    if (!this.noteEntriesById.has(noteId)) {
+      return false;
+    }
+
+    this.removeNoteSnapshot(noteId);
+    this.persistKnownWordCacheState();
+    log.info('Known-word cache removed deleted note', `noteId=${noteId}`);
+    return true;
+  }
+
  clearKnownWordCacheState(): void {
    this.clearInMemoryState();
    this.knownWordsStateKey = this.getKnownWordCacheStateKey();
@@ -44,6 +44,7 @@ function createWorkflowHarness() {
        updates.push({ noteId, fields });
      },
      storeMediaFile: async () => undefined,
+      deleteNotes: async () => undefined,
    },
    getConfig: () => ({
      fields: {
@@ -58,9 +59,10 @@ function createWorkflowHarness() {
      sentenceField: 'Sentence',
      lapisEnabled: false,
      kikuEnabled: false,
-      kikuFieldGrouping: 'disabled' as const,
+      fieldGroupingMode: 'disabled' as const,
    }),
    appendKnownWordsFromNoteInfo: (_noteInfo: NoteUpdateWorkflowNoteInfo) => undefined,
+    removeKnownWordNote: (_noteId: number) => undefined,
    extractFields: (fields: Record<string, { value: string }>) => {
      const out: Record<string, string> = {};
      for (const [key, value] of Object.entries(fields)) {
@@ -80,7 +82,6 @@ function createWorkflowHarness() {
      const names = Object.keys(noteInfo.fields);
      return names.find((name) => name.toLowerCase() === preferred.toLowerCase()) ?? null;
    },
-    getResolvedSentenceAudioFieldName: () => null,
    getAnimatedImageLeadInSeconds: async () => 0,
    mergeFieldValue: (_existing: string, next: string, _overwrite: boolean) => next,
    generateAudioFilename: () => 'audio_1.mp3',
@@ -120,6 +121,49 @@ test('NoteUpdateWorkflow updates sentence field and emits notification', async (
  assert.equal(harness.notifications.length, 1);
 });

+test('NoteUpdateWorkflow uses configured fields for word-card enrichment with Lapis and Kiku enabled', async () => {
+  const harness = createWorkflowHarness();
+  harness.deps.getConfig = () => ({
+    fields: {
+      sentence: 'Context',
+      audio: 'ContextAudio',
+    },
+    media: {
+      generateAudio: true,
+      generateImage: false,
+    },
+    behavior: {},
+  });
+  harness.deps.getEffectiveSentenceCardConfig = () => ({
+    sentenceField: 'Sentence',
+    lapisEnabled: true,
+    kikuEnabled: true,
+    fieldGroupingMode: 'disabled',
+  });
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'taberu' },
+          Sentence: { value: '' },
+          SentenceAudio: { value: '' },
+          Context: { value: '' },
+          ContextAudio: { value: '' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+  harness.deps.generateAudio = async () => Buffer.from('audio');
+
+  await harness.workflow.execute(42);
+
+  assert.equal(harness.updates.length, 1);
+  assert.deepEqual(harness.updates[0]?.fields, {
+    Context: 'subtitle-text',
+    ContextAudio: '[sound:audio_1.mp3]',
+  });
+});
+
 test('NoteUpdateWorkflow updates sentence furigana when highlight processor changes it', async () => {
  const harness = createWorkflowHarness();
  harness.deps.client.notesInfo = async () =>
@@ -151,7 +195,7 @@ test('NoteUpdateWorkflow marks enriched Kiku word cards as word-and-sentence car
    sentenceField: 'Sentence',
    lapisEnabled: false,
    kikuEnabled: true,
-    kikuFieldGrouping: 'manual',
+    fieldGroupingMode: 'manual',
  });
  harness.deps.client.notesInfo = async () =>
    [
@@ -184,7 +228,7 @@ test('NoteUpdateWorkflow marks the configured word card kind instead of word-and
    sentenceField: 'Sentence',
    lapisEnabled: false,
    kikuEnabled: true,
-    kikuFieldGrouping: 'manual',
+    fieldGroupingMode: 'manual',
    wordCardKind: 'click',
  });
  harness.deps.client.notesInfo = async () =>
@@ -220,7 +264,7 @@ test('NoteUpdateWorkflow leaves card type flags alone when the word card kind is
    sentenceField: 'Sentence',
    lapisEnabled: false,
    kikuEnabled: true,
-    kikuFieldGrouping: 'manual',
+    fieldGroupingMode: 'manual',
    wordCardKind: 'none',
  });
  harness.deps.client.notesInfo = async () =>
@@ -275,7 +319,7 @@ test('NoteUpdateWorkflow preserves explicit sentence card type during sentence e
    sentenceField: 'Sentence',
    lapisEnabled: true,
    kikuEnabled: false,
-    kikuFieldGrouping: 'disabled',
+    fieldGroupingMode: 'disabled',
  });
  harness.deps.client.notesInfo = async () =>
    [
@@ -318,7 +362,7 @@ test('NoteUpdateWorkflow updates note before auto field grouping merge', async (
    sentenceField: 'Sentence',
    lapisEnabled: false,
    kikuEnabled: true,
-    kikuFieldGrouping: 'auto',
+    fieldGroupingMode: 'auto',
  });
  harness.deps.findDuplicateNote = async () => 99;
  harness.deps.client.notesInfo = async () => {
@@ -432,6 +476,7 @@ test('NoteUpdateWorkflow uses subtitle sidebar context for sentence media timing
  harness.deps.getConfig = () => ({
    fields: {
      sentence: 'Sentence',
+      audio: 'SentenceAudio',
      image: 'Picture',
      miscInfo: 'MiscInfo',
    },
@@ -444,7 +489,6 @@ test('NoteUpdateWorkflow uses subtitle sidebar context for sentence media timing
  });
  harness.deps.getCurrentSubtitleText = () => 'current primary line';
  harness.deps.getCurrentSubtitleStart = () => 20;
-  harness.deps.getResolvedSentenceAudioFieldName = () => 'SentenceAudio';
  harness.deps.generateAudio = async (context?: SubtitleMiningContext) => {
    audioContext = context ?? null;
    return Buffer.from('audio');
@@ -501,6 +545,7 @@ test('NoteUpdateWorkflow snapshots one media range for audio and image without a
  harness.deps.getConfig = () => ({
    fields: {
      sentence: 'Sentence',
+      audio: 'SentenceAudio',
      image: 'Picture',
      miscInfo: 'MiscInfo',
    },
@@ -511,7 +556,6 @@ test('NoteUpdateWorkflow snapshots one media range for audio and image without a
    },
    behavior: {},
  });
-  harness.deps.getResolvedSentenceAudioFieldName = () => 'SentenceAudio';
  harness.deps.captureSubtitleMediaContext = () => {
    captureCalls += 1;
    return capturedContext;
@@ -592,3 +636,141 @@ test('NoteUpdateWorkflow queues media updates when YouTube cache is pending', as
  assert.equal(queuedUpdates[0]?.context, undefined);
  assert.deepEqual(harness.updates, [{ noteId: 42, fields: { Sentence: 'subtitle-text' } }]);
 });
+
+test('NoteUpdateWorkflow deletes an existing word card when timing review discards it', async () => {
+  const harness = createWorkflowHarness();
+  const deletedNoteIds: number[][] = [];
+  const removedKnownWordNoteIds: number[] = [];
+  let appendedKnownWords = false;
+  harness.deps.captureSubtitleMediaContext = () => ({
+    source: 'overlay',
+    text: 'subtitle-text',
+    startTime: 4,
+    endTime: 6,
+  });
+  harness.deps.client.deleteNotes = async (noteIds) => {
+    deletedNoteIds.push(noteIds);
+  };
+  harness.deps.appendKnownWordsFromNoteInfo = () => {
+    appendedKnownWords = true;
+  };
+  harness.deps.removeKnownWordNote = (noteId) => {
+    removedKnownWordNoteIds.push(noteId);
+  };
+  harness.deps.reviewMediaTiming = async () => ({ action: 'discard' });
+
+  await harness.workflow.execute(42);
+
+  assert.deepEqual(deletedNoteIds, [[42]]);
+  assert.deepEqual(removedKnownWordNoteIds, [42]);
+  assert.equal(appendedKnownWords, false);
+  assert.deepEqual(harness.updates, []);
+  assert.deepEqual(harness.notifications, []);
+});
+
+test('NoteUpdateWorkflow keeps the word card but skips media after timing review', async () => {
+  const harness = createWorkflowHarness();
+  const mediaCalls: string[] = [];
+  const deletedNoteIds: number[][] = [];
+  const queuedUpdates: unknown[] = [];
+  harness.deps.captureSubtitleMediaContext = () => ({
+    source: 'overlay',
+    text: 'subtitle-text',
+    startTime: 4,
+    endTime: 6,
+  });
+  harness.deps.getConfig = () => ({
+    fields: { sentence: 'Sentence', image: 'Picture' },
+    media: { generateAudio: true, generateImage: true },
+    behavior: {},
+  });
+  harness.deps.reviewMediaTiming = async () => ({ action: 'skip-media' });
+  harness.deps.generateAudio = async () => {
+    mediaCalls.push('audio');
+    return Buffer.from('audio');
+  };
+  harness.deps.generateImage = async () => {
+    mediaCalls.push('image');
+    return Buffer.from('image');
+  };
+  harness.deps.queuePendingYoutubeMediaUpdate = async (update) => {
+    queuedUpdates.push(update);
+    return true;
+  };
+  harness.deps.client.deleteNotes = async (noteIds) => {
+    deletedNoteIds.push(noteIds);
+  };
+
+  await harness.workflow.execute(42);
+
+  assert.deepEqual(mediaCalls, []);
+  assert.deepEqual(queuedUpdates, []);
+  assert.deepEqual(deletedNoteIds, []);
+  assert.deepEqual(harness.updates, [{ noteId: 42, fields: { Sentence: 'subtitle-text' } }]);
+  assert.deepEqual(harness.notifications, [{ noteId: 42, label: 'taberu' }]);
+});
+
+test('NoteUpdateWorkflow uses the combined review sentence for the card and media range', async () => {
+  const harness = createWorkflowHarness();
+  const audioContexts: Array<SubtitleMiningContext | undefined> = [];
+  harness.deps.captureSubtitleMediaContext = () => ({
+    source: 'overlay',
+    text: 'current-line',
+    startTime: 4,
+    endTime: 6,
+  });
+  harness.deps.getConfig = () => ({
+    fields: { sentence: 'Sentence' },
+    media: { generateAudio: true, generateImage: false },
+    behavior: {},
+  });
+  harness.deps.reviewMediaTiming = async () => ({
+    action: 'confirm',
+    startTime: 2,
+    endTime: 7,
+    text: 'previous-line current-line next-line',
+  });
+  harness.deps.generateAudio = async (context) => {
+    audioContexts.push(context);
+    return null;
+  };
+
+  await harness.workflow.execute(42);
+
+  assert.deepEqual(harness.updates, [
+    { noteId: 42, fields: { Sentence: 'previous-line current-line next-line' } },
+  ]);
+  assert.equal(audioContexts.length, 1);
+  assert.equal(audioContexts[0]?.text, 'previous-line current-line next-line');
+  assert.equal(audioContexts[0]?.startTime, 2);
+  assert.equal(audioContexts[0]?.endTime, 7);
+  assert.equal(audioContexts[0]?.mediaPaddingSeconds, 0);
+});
+
+test('NoteUpdateWorkflow keeps cache unchanged and reports when deletion fails', async () => {
+  const harness = createWorkflowHarness();
+  const statusMessages: string[] = [];
+  let removedKnownWord = false;
+  harness.deps.captureSubtitleMediaContext = () => ({
+    source: 'overlay',
+    text: 'subtitle-text',
+    startTime: 4,
+    endTime: 6,
+  });
+  harness.deps.client.deleteNotes = async () => {
+    throw new Error('delete failed');
+  };
+  harness.deps.removeKnownWordNote = () => {
+    removedKnownWord = true;
+  };
+  harness.deps.showOsdNotification = (message) => {
+    statusMessages.push(message);
+  };
+  harness.deps.reviewMediaTiming = async () => ({ action: 'discard' });
+
+  await harness.workflow.execute(42);
+
+  assert.equal(removedKnownWord, false);
+  assert.deepEqual(statusMessages, ['Card deletion failed: delete failed']);
+  assert.ok(harness.warnings.length === 0);
+});
@@ -1,7 +1,12 @@
 import { DEFAULT_ANKI_CONNECT_CONFIG } from '../config';
 import { getPreferredWordValueFromExtractedFields } from '../anki-field-config';
 import type { SubtitleMiningContext } from '../types/subtitle';
-import type { CardKind, WordCardKind } from '../types/anki';
+import type {
+  CardKind,
+  MediaTimingReviewDecision,
+  MediaTimingReviewRequest,
+  WordCardKind,
+} from '../types/anki';
 import { resolveWordCardKind } from './note-field-utils';

 export interface NoteUpdateWorkflowNoteInfo {
@@ -14,11 +19,13 @@ export interface NoteUpdateWorkflowDeps {
    notesInfo(noteIds: number[]): Promise<unknown>;
    updateNoteFields(noteId: number, fields: Record<string, string>): Promise<void>;
    storeMediaFile(filename: string, data: Buffer): Promise<void>;
+    deleteNotes(noteIds: number[]): Promise<void>;
  };
  getConfig: () => {
    fields?: {
      word?: string;
      sentence?: string;
+      audio?: string;
      image?: string;
      miscInfo?: string;
    };
@@ -39,10 +46,11 @@ export interface NoteUpdateWorkflowDeps {
    sentenceField: string;
    lapisEnabled: boolean;
    kikuEnabled: boolean;
-    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
    wordCardKind?: WordCardKind;
  };
  appendKnownWordsFromNoteInfo: (noteInfo: NoteUpdateWorkflowNoteInfo) => void;
+  removeKnownWordNote: (noteId: number) => void;
  extractFields: (fields: Record<string, { value: string }>) => Record<string, string>;
  findDuplicateNote: (
    expression: string,
@@ -75,7 +83,6 @@ export interface NoteUpdateWorkflowDeps {
    noteInfo: NoteUpdateWorkflowNoteInfo,
    ...preferredNames: (string | undefined)[]
  ) => string | null;
-  getResolvedSentenceAudioFieldName: (noteInfo: NoteUpdateWorkflowNoteInfo) => string | null;
  getAnimatedImageLeadInSeconds: (noteInfo: NoteUpdateWorkflowNoteInfo) => Promise<number>;
  mergeFieldValue: (existing: string, newValue: string, overwrite: boolean) => string;
  generateAudioFilename: () => string;
@@ -102,6 +109,9 @@ export interface NoteUpdateWorkflowDeps {
  logWarn: (message: string, ...args: unknown[]) => void;
  logInfo: (message: string, ...args: unknown[]) => void;
  logError: (message: string, ...args: unknown[]) => void;
+  reviewMediaTiming?: (
+    request: Omit<MediaTimingReviewRequest, 'audioPadding' | 'maxMediaDuration'>,
+  ) => Promise<MediaTimingReviewDecision>;
 }

 function normalizeSubtitleContextText(text: string): string {
@@ -160,7 +170,7 @@ export class NoteUpdateWorkflow {
    return null;
  }

-  async execute(noteId: number, options?: { skipKikuFieldGrouping?: boolean }): Promise<void> {
+  async execute(noteId: number, options?: { skipFieldGrouping?: boolean }): Promise<void> {
    this.deps.beginUpdateProgress('Updating card');
    try {
      const notesInfoResult = await this.deps.client.notesInfo([noteId]);
@@ -171,7 +181,6 @@ export class NoteUpdateWorkflow {
      }

      const noteInfo = notesInfo[0]!;
-      this.deps.appendKnownWordsFromNoteInfo(noteInfo);
      const fields = this.deps.extractFields(noteInfo.fields);
      const config = this.deps.getConfig();

@@ -187,9 +196,7 @@ export class NoteUpdateWorkflow {

      const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
      const shouldRunFieldGrouping =
-        !options?.skipKikuFieldGrouping &&
-        sentenceCardConfig.kikuEnabled &&
-        sentenceCardConfig.kikuFieldGrouping !== 'disabled';
+        !options?.skipFieldGrouping && sentenceCardConfig.fieldGroupingMode !== 'disabled';
      let duplicateNoteId: number | null = null;
      if (shouldRunFieldGrouping && hasExpressionText) {
        duplicateNoteId = await this.deps.findDuplicateNote(expressionText, noteId, noteInfo);
@@ -198,20 +205,64 @@ export class NoteUpdateWorkflow {
      const updatedFields: Record<string, string> = {};
      let updatePerformed = false;
      let miscInfoFilename: string | null = null;
-      const sentenceField = sentenceCardConfig.sentenceField;
+      const configuredSentenceField =
+        config.fields?.sentence ?? DEFAULT_ANKI_CONNECT_CONFIG.fields.sentence;
+      const sentenceField = this.deps.resolveConfiguredFieldName(noteInfo, configuredSentenceField);
      const subtitleMiningContext = this.consumeMatchingSubtitleMiningContext(
        fields,
-        sentenceField,
-        config.fields?.sentence,
+        sentenceField ?? configuredSentenceField,
+        configuredSentenceField,
      );
      // Audio and image generation run sequentially and audio extraction can take tens of
      // seconds, so resolve the clip range exactly once up front; reading live mpv sub
      // timings per generator clips whichever line is on screen when each one starts.
-      const mediaTimingContext =
+      let mediaTimingContext =
        subtitleMiningContext ?? this.deps.captureSubtitleMediaContext?.() ?? null;
+      let skipMedia = false;
+      let reviewedSentenceText: string | undefined;
      const noteLabel = hasExpressionText ? expressionText : noteId;

-      const currentSubtitleText = subtitleMiningContext?.text ?? this.deps.getCurrentSubtitleText();
+      if (mediaTimingContext) {
+        const timingDecision = this.deps.reviewMediaTiming
+          ? await this.deps.reviewMediaTiming({
+              kind: 'word',
+              text: mediaTimingContext.text,
+              startTime: mediaTimingContext.startTime,
+              endTime: mediaTimingContext.endTime,
+              noteId,
+            })
+          : ({ action: 'use-original' } as const);
+        if (timingDecision.action === 'discard') {
+          try {
+            await this.deps.client.deleteNotes([noteId]);
+          } catch (error) {
+            const message = error instanceof Error ? error.message : String(error);
+            this.deps.logError('Failed to delete discarded card:', message);
+            this.deps.showOsdNotification(`Card deletion failed: ${message}`);
+            return;
+          }
+          this.deps.removeKnownWordNote(noteId);
+          this.deps.showOsdNotification('Card deleted.');
+          return;
+        }
+        if (timingDecision.action === 'confirm') {
+          reviewedSentenceText = timingDecision.text?.trim() || undefined;
+          mediaTimingContext = {
+            ...mediaTimingContext,
+            ...(reviewedSentenceText !== undefined ? { text: reviewedSentenceText } : {}),
+            startTime: timingDecision.startTime,
+            endTime: timingDecision.endTime,
+            mediaPaddingSeconds: 0,
+          };
+        } else if (timingDecision.action === 'skip-media') {
+          skipMedia = true;
+        }
+      }
+
+      this.deps.appendKnownWordsFromNoteInfo(noteInfo);
+
+      const currentSubtitleText =
+        reviewedSentenceText ?? subtitleMiningContext?.text ?? this.deps.getCurrentSubtitleText();
      if (sentenceField && currentSubtitleText) {
        const processedSentence = this.deps.processSentence(currentSubtitleText, fields);
        updatedFields[sentenceField] = processedSentence;
@@ -239,8 +290,8 @@ export class NoteUpdateWorkflow {
        }
      }

-      const generateAudio = config.media?.generateAudio !== false;
-      const generateImage = config.media?.generateImage !== false;
+      const generateAudio = !skipMedia && config.media?.generateAudio !== false;
+      const generateImage = !skipMedia && config.media?.generateImage !== false;
      const mediaCacheQueued =
        (generateAudio || generateImage) && this.deps.queuePendingYoutubeMediaUpdate
          ? await this.deps.queuePendingYoutubeMediaUpdate({
@@ -258,7 +309,10 @@ export class NoteUpdateWorkflow {

          if (audioBuffer) {
            await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
-            const sentenceAudioField = this.deps.getResolvedSentenceAudioFieldName(noteInfo);
+            const sentenceAudioField = this.deps.resolveConfiguredFieldName(
+              noteInfo,
+              config.fields?.audio ?? DEFAULT_ANKI_CONNECT_CONFIG.fields.audio,
+            );
            if (sentenceAudioField) {
              const existingAudio = noteInfo.fields[sentenceAudioField]?.value || '';
              updatedFields[sentenceAudioField] = this.deps.mergeFieldValue(
@@ -345,7 +399,7 @@ export class NoteUpdateWorkflow {
          noteInfoForGrouping = refreshedInfo[0]!;
        }

-        if (sentenceCardConfig.kikuFieldGrouping === 'auto') {
+        if (sentenceCardConfig.fieldGroupingMode === 'auto') {
          await this.deps.handleFieldGroupingAuto(
            duplicateNoteId,
            noteId,
@@ -354,7 +408,7 @@ export class NoteUpdateWorkflow {
          );
          return;
        }
-        if (sentenceCardConfig.kikuFieldGrouping === 'manual') {
+        if (sentenceCardConfig.fieldGroupingMode === 'manual') {
          await this.deps.handleFieldGroupingManual(
            duplicateNoteId,
            noteId,
@@ -31,7 +31,6 @@ function createDeps(
    getCachedMediaPath: async () => null,
    shouldRequireRemoteMediaCache: () => true,
    getSubtitleMediaRange: () => ({ startTime: 1, endTime: 2 }),
-    getResolvedSentenceAudioFieldName: () => 'SentenceAudio',
    resolveConfiguredFieldName: () => 'Picture',
    mergeFieldValue: (_existing, newValue) => newValue,
    getAnimatedImageLeadInSeconds: async () => 0,
@@ -133,7 +132,7 @@ test('PendingYoutubeMediaQueue defaults missing media flags to enabled when queu
        noteIds.map((noteId) => ({
          noteId,
          fields: {
-            SentenceAudio: { value: '' },
+            ExpressionAudio: { value: '' },
            Picture: { value: '' },
          },
        })),
@@ -144,13 +143,16 @@ test('PendingYoutubeMediaQueue defaults missing media flags to enabled when queu
        storedMedia.push(filename);
      },
    },
-    getConfig: () => ({ media: {}, fields: { image: 'Picture' } }) as AnkiConnectConfig,
+    getConfig: () =>
+      ({ media: {}, fields: { audio: 'ExpressionAudio', image: 'Picture' } }) as AnkiConnectConfig,
+    resolveConfiguredFieldName: (noteInfo, ...preferredNames) =>
+      preferredNames.find((name) => name && name in noteInfo.fields) ?? null,
  });
  const queue = new PendingYoutubeMediaQueue(deps);

  const queued = await queue.queueFromNote({
    noteId: 42,
-    noteInfo: { noteId: 42, fields: {} },
+    noteInfo: { noteId: 42, fields: { ExpressionAudio: { value: '' } } },
    label: 'demo',
  });
  await queue.handleReady('https://youtu.be/abc123', '/tmp/media.mkv');
@@ -158,7 +160,8 @@ test('PendingYoutubeMediaQueue defaults missing media flags to enabled when queu
  assert.equal(queued, true);
  assert.equal(updatedNotes.length, 1);
  assert.equal(storedMedia.length, 2);
-  assert.match(updatedNotes[0]?.fields.SentenceAudio ?? '', /^\[sound:audio\.mp3\]$/);
+  assert.match(updatedNotes[0]?.fields.ExpressionAudio ?? '', /^\[sound:audio\.mp3\]$/);
+  assert.equal('SentenceAudio' in (updatedNotes[0]?.fields ?? {}), false);
  assert.match(updatedNotes[0]?.fields.Picture ?? '', /^<img src="image\.webp">$/);
 });

@@ -39,7 +39,6 @@ export interface PendingYoutubeMediaQueueDeps {
    startTime: number;
    endTime: number;
  };
-  getResolvedSentenceAudioFieldName: (noteInfo: PendingYoutubeMediaNoteInfo) => string | null;
  resolveConfiguredFieldName: (
    noteInfo: PendingYoutubeMediaNoteInfo,
    ...preferredNames: (string | undefined)[]
@@ -136,7 +135,7 @@ export class PendingYoutubeMediaQueue {
      startTime: mediaRange.startTime,
      endTime: mediaRange.endTime,
      label: job.label,
-      audioFieldName: this.deps.getResolvedSentenceAudioFieldName(job.noteInfo) ?? undefined,
+      audioFieldName: this.resolveConfiguredAudioFieldName(job.noteInfo) ?? undefined,
      imageFieldName:
        this.deps.resolveConfiguredFieldName(
          job.noteInfo,
@@ -148,6 +147,9 @@ export class PendingYoutubeMediaQueue {
      generateAudio: shouldGenerateAudio(config),
      generateImage: shouldGenerateImage(config),
      volumeScale,
+      ...(job.context?.mediaPaddingSeconds !== undefined
+        ? { mediaPaddingSeconds: job.context.mediaPaddingSeconds }
+        : {}),
    });
    return true;
  }
@@ -247,6 +249,14 @@ export class PendingYoutubeMediaQueue {
    return matched;
  }

+  private resolveConfiguredAudioFieldName(noteInfo: PendingYoutubeMediaNoteInfo): string | null {
+    const config = this.deps.getConfig();
+    return this.deps.resolveConfiguredFieldName(
+      noteInfo,
+      config.fields?.audio ?? DEFAULT_ANKI_CONNECT_CONFIG.fields.audio,
+    );
+  }
+
  private async applyUpdate(
    job: PendingYoutubeMediaUpdate,
    cachedPath: string,
@@ -275,7 +285,7 @@ export class PendingYoutubeMediaQueue {
          cachedMediaInput,
          job.startTime,
          job.endTime,
-          config.media?.audioPadding,
+          job.mediaPaddingSeconds ?? config.media?.audioPadding,
          undefined,
          config.media?.normalizeAudio !== false,
          job.volumeScale,
@@ -283,7 +293,7 @@ export class PendingYoutubeMediaQueue {
        if (audioBuffer) {
          await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
          const audioField =
-            job.audioFieldName || this.deps.getResolvedSentenceAudioFieldName(noteInfo) || null;
+            job.audioFieldName || this.resolveConfiguredAudioFieldName(noteInfo) || null;
          if (audioField) {
            const existingAudio = noteInfo.fields[audioField]?.value || '';
            mediaFields[audioField] = this.deps.mergeFieldValue(
@@ -309,6 +319,7 @@ export class PendingYoutubeMediaQueue {
          job.startTime,
          job.endTime,
          animatedLeadInSeconds,
+          job.mediaPaddingSeconds,
        );
        if (imageBuffer) {
          await this.deps.client.storeMediaFile(imageFilename, imageBuffer);
@@ -369,6 +380,7 @@ export class PendingYoutubeMediaQueue {
    startTime: number,
    endTime: number,
    animatedLeadInSeconds = 0,
+    mediaPaddingSeconds?: number,
  ): Promise<Buffer | null> {
    const config = this.deps.getConfig();
    if (config.media?.imageType === 'avif') {
@@ -376,7 +388,7 @@ export class PendingYoutubeMediaQueue {
        videoPath,
        startTime,
        endTime,
-        config.media?.audioPadding,
+        mediaPaddingSeconds ?? config.media?.audioPadding,
        {
          fps: config.media?.animatedFps,
          maxWidth: config.media?.animatedMaxWidth,
@@ -10,6 +10,7 @@ export interface PendingYoutubeMediaUpdate {
  generateAudio: boolean;
  generateImage: boolean;
  volumeScale?: number;
+  mediaPaddingSeconds?: number;
 }

 function trimToNonEmptyString(value: unknown): string | null {
@@ -116,6 +116,10 @@ export function normalizeAnkiIntegrationConfig(config: AnkiConnectConfig): AnkiC
      ...DEFAULT_ANKI_CONNECT_CONFIG.isKiku,
      ...(config.isKiku ?? {}),
    },
+    isSenren: {
+      ...DEFAULT_ANKI_CONNECT_CONFIG.isSenren,
+      ...(config.isSenren ?? {}),
+    },
    lapisKiku: {
      ...DEFAULT_ANKI_CONNECT_CONFIG.lapisKiku,
      ...(config.lapisKiku ?? {}),
@@ -209,6 +213,10 @@ export class AnkiIntegrationRuntime {
        patch.isKiku !== undefined
          ? { ...this.config.isKiku, ...patch.isKiku }
          : this.config.isKiku,
+      isSenren:
+        patch.isSenren !== undefined
+          ? { ...this.config.isSenren, ...patch.isSenren }
+          : this.config.isSenren,
      lapisKiku:
        patch.lapisKiku !== undefined
          ? { ...this.config.lapisKiku, ...patch.lapisKiku }
@@ -2181,6 +2181,7 @@ test('runtime options registry is centralized', () => {
  const ids = RUNTIME_OPTION_REGISTRY.map((entry) => entry.id);
  assert.deepEqual(ids, [
    'anki.autoUpdateNewCards',
+    'anki.mediaReviewTiming',
    'subtitle.annotation.knownWords.highlightEnabled',
    'subtitle.annotation.knownWords.maturityEnabled',
    'subtitle.annotation.nPlusOne',
@@ -2188,6 +2189,7 @@ test('runtime options registry is centralized', () => {
    'subtitle.annotation.frequency',
    'anki.nPlusOneMatchMode',
    'anki.kikuFieldGrouping',
+    'anki.senrenFieldGrouping',
  ]);
 });

@@ -2775,6 +2777,47 @@ test('accepts a Kiku/Lapis word card kind and warns on an unknown one', () => {
  );
 });

+test('forces Senren off when Kiku is also enabled and validates Senren fieldGrouping', () => {
+  const dir = makeTempDir();
+  fs.writeFileSync(
+    path.join(dir, 'config.jsonc'),
+    `{
+      "ankiConnect": {
+        "isKiku": { "enabled": true },
+        "isSenren": { "enabled": true }
+      }
+    }`,
+    'utf-8',
+  );
+
+  const service = new ConfigService(dir);
+  assert.equal(service.getConfig().ankiConnect.isKiku.enabled, true);
+  assert.equal(service.getConfig().ankiConnect.isSenren.enabled, false);
+  assert.ok(
+    service.getWarnings().some((warning) => warning.path === 'ankiConnect.isSenren.enabled'),
+  );
+
+  const senrenOnlyDir = makeTempDir();
+  fs.writeFileSync(
+    path.join(senrenOnlyDir, 'config.jsonc'),
+    `{
+      "ankiConnect": {
+        "isSenren": { "enabled": true, "fieldGrouping": "sometimes" }
+      }
+    }`,
+    'utf-8',
+  );
+
+  const senrenOnlyService = new ConfigService(senrenOnlyDir);
+  assert.equal(senrenOnlyService.getConfig().ankiConnect.isSenren.enabled, true);
+  assert.equal(senrenOnlyService.getConfig().ankiConnect.isSenren.fieldGrouping, 'auto');
+  assert.ok(
+    senrenOnlyService
+      .getWarnings()
+      .some((warning) => warning.path === 'ankiConnect.isSenren.fieldGrouping'),
+  );
+});
+
 test('accepts valid ankiConnect knownWords deck object', () => {
  const dir = makeTempDir();
  fs.writeFileSync(
@@ -54,6 +54,7 @@ export const INTEGRATIONS_DEFAULT_CONFIG: Pick<
      syncAnimatedImageToWordAudio: true,
      normalizeAudio: true,
      mirrorMpvVolume: true,
+      reviewTiming: false,
      audioPadding: 0,
      fallbackDuration: 3.0,
      maxMediaDuration: 30,
@@ -91,6 +92,11 @@ export const INTEGRATIONS_DEFAULT_CONFIG: Pick<
      fieldGrouping: 'disabled',
      deleteDuplicateInAuto: true,
    },
+    isSenren: {
+      enabled: false,
+      fieldGrouping: 'auto',
+      deleteDuplicateInAuto: true,
+    },
    lapisKiku: {
      wordCardKind: 'word-and-sentence',
    },
@@ -196,6 +196,14 @@ export function buildIntegrationConfigOptionRegistry(
      description:
        "Apply mpv's current software volume curve to generated sentence audio. Changes apply live.",
    },
+    {
+      path: 'ankiConnect.media.reviewTiming',
+      kind: 'boolean',
+      defaultValue: defaultConfig.ankiConnect.media.reviewTiming,
+      description:
+        'Review and preview subtitle media timing before SubMiner creates or enriches a mined card.',
+      runtime: runtimeOptionById.get('anki.mediaReviewTiming'),
+    },
    {
      path: 'ankiConnect.media.generateImage',
      kind: 'boolean',
@@ -363,6 +371,28 @@ export function buildIntegrationConfigOptionRegistry(
      description:
        'When Kiku field grouping is "auto", delete the duplicate source card after grouping completes.',
    },
+    {
+      path: 'ankiConnect.isSenren.fieldGrouping',
+      kind: 'enum',
+      enumValues: ['auto', 'manual', 'disabled'],
+      defaultValue: defaultConfig.ankiConnect.isSenren.fieldGrouping,
+      description: 'Senren duplicate-card field grouping mode (scene switching).',
+      runtime: runtimeOptionById.get('anki.senrenFieldGrouping'),
+    },
+    {
+      path: 'ankiConnect.isSenren.enabled',
+      kind: 'boolean',
+      defaultValue: defaultConfig.ankiConnect.isSenren.enabled,
+      description:
+        'Enable Senren-specific duplicate handling (scene-switching field grouping, including miscInfo grouping). Mutually exclusive with isKiku.enabled.',
+    },
+    {
+      path: 'ankiConnect.isSenren.deleteDuplicateInAuto',
+      kind: 'boolean',
+      defaultValue: defaultConfig.ankiConnect.isSenren.deleteDuplicateInAuto,
+      description:
+        'When Senren field grouping is "auto", delete the duplicate source card after grouping completes.',
+    },
    {
      path: 'ankiConnect.isLapis.enabled',
      kind: 'boolean',
@@ -19,6 +19,20 @@ export function buildRuntimeOptionRegistry(
        behavior: { autoUpdateNewCards: value === true },
      }),
    },
+    {
+      id: 'anki.mediaReviewTiming',
+      path: 'ankiConnect.media.reviewTiming',
+      label: 'Review Media Timing',
+      scope: 'ankiConnect',
+      valueType: 'boolean',
+      allowedValues: [true, false],
+      defaultValue: defaultConfig.ankiConnect.media.reviewTiming,
+      requiresRestart: false,
+      formatValueForOsd: (value) => (value === true ? 'On' : 'Off'),
+      toAnkiPatch: (value) => ({
+        media: { reviewTiming: value === true },
+      }),
+    },
    {
      id: 'subtitle.annotation.knownWords.highlightEnabled',
      path: 'ankiConnect.knownWords.highlightEnabled',
@@ -124,5 +138,22 @@ export function buildRuntimeOptionRegistry(
        },
      }),
    },
+    {
+      id: 'anki.senrenFieldGrouping',
+      path: 'ankiConnect.isSenren.fieldGrouping',
+      label: 'Senren Field Grouping',
+      scope: 'ankiConnect',
+      valueType: 'enum',
+      allowedValues: ['auto', 'manual', 'disabled'],
+      defaultValue: 'auto',
+      requiresRestart: false,
+      formatValueForOsd: (value) => String(value),
+      toAnkiPatch: (value) => ({
+        isSenren: {
+          fieldGrouping:
+            value === 'auto' || value === 'manual' || value === 'disabled' ? value : 'auto',
+        },
+      }),
+    },
  ];
 }
@@ -135,7 +135,7 @@ const INTEGRATION_TEMPLATE_SECTIONS: ConfigTemplateSection[] = [
    title: 'AnkiConnect Integration',
    description: ['Automatic Anki updates and media generation options.'],
    notes: [
-      'Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.',
+      'Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume/reviewTiming, knownWords, nPlusOne, fields.word/audio/image/sentence/miscInfo, behavior.autoUpdateNewCards, isLapis.sentenceCardModel, isKiku.fieldGrouping, isSenren.fieldGrouping, and lapisKiku.wordCardKind update live while SubMiner is running.',
      'Shared AI provider transport settings are read from top-level ai and typically require restart.',
      'Most other AnkiConnect settings still require restart.',
    ],
@@ -21,6 +21,34 @@ function makeContext(ankiConnect: unknown): {
  return { context, warnings };
 }

+test('media timing review is disabled by default and accepts a boolean override', () => {
+  const defaultContext = makeContext({});
+  applyAnkiConnectResolution(defaultContext.context);
+  assert.equal(defaultContext.context.resolved.ankiConnect.media.reviewTiming, false);
+
+  const enabledContext = makeContext({ media: { reviewTiming: true } });
+  applyAnkiConnectResolution(enabledContext.context);
+  assert.equal(enabledContext.context.resolved.ankiConnect.media.reviewTiming, true);
+  assert.deepEqual(enabledContext.warnings, []);
+});
+
+test('modern media duration accepts zero as the disabled cap sentinel', () => {
+  const disabledCap = makeContext({ media: { maxMediaDuration: 0 } });
+  applyAnkiConnectResolution(disabledCap.context);
+  assert.equal(disabledCap.context.resolved.ankiConnect.media.maxMediaDuration, 0);
+  assert.deepEqual(disabledCap.warnings, []);
+
+  const invalidCap = makeContext({ media: { maxMediaDuration: -1 } });
+  applyAnkiConnectResolution(invalidCap.context);
+  assert.equal(
+    invalidCap.context.resolved.ankiConnect.media.maxMediaDuration,
+    DEFAULT_CONFIG.ankiConnect.media.maxMediaDuration,
+  );
+  assert.ok(
+    invalidCap.warnings.some((warning) => warning.path === 'ankiConnect.media.maxMediaDuration'),
+  );
+});
+
 test('modern invalid knownWords.highlightEnabled warns modern key and does not fallback to legacy', () => {
  const { context, warnings } = makeContext({
    nPlusOne: { highlightEnabled: true },
@@ -1,6 +1,7 @@
 import type { ResolveContext } from './context';
 import { initializeAnkiConnectResolution } from './anki-connect/initialize';
 import { applyAnkiKikuResolution } from './anki-connect/kiku';
+import { applyAnkiSenrenResolution } from './anki-connect/senren';
 import { applyAnkiLapisKikuResolution } from './anki-connect/lapis-kiku';
 import { applyAnkiKnownWordsResolution } from './anki-connect/known-words';
 import { applyAnkiLegacyResolution } from './anki-connect/legacy';
@@ -23,5 +24,6 @@ export function applyAnkiConnectResolution(context: ResolveContext): void {
  applyAnkiLegacyResolution(context, ankiConnect, behavior, fields, media, metadata);
  applyAnkiKnownWordsResolution(context, ankiConnect, behavior);
  applyAnkiKikuResolution(context);
+  applyAnkiSenrenResolution(context);
  applyAnkiLapisKikuResolution(context, ankiConnect);
 }
@@ -77,6 +77,12 @@ export function initializeAnkiConnectResolution(
        ? (ankiConnect.isKiku as (typeof context.resolved)['ankiConnect']['isKiku'])
        : {}),
    },
+    isSenren: {
+      ...context.resolved.ankiConnect.isSenren,
+      ...(isObject(ankiConnect.isSenren)
+        ? (ankiConnect.isSenren as (typeof context.resolved)['ankiConnect']['isSenren'])
+        : {}),
+    },
    lapisKiku: {
      ...context.resolved.ankiConnect.lapisKiku,
    },
@@ -19,6 +19,7 @@ export function applyModernMediaResolution(
    'syncAnimatedImageToWordAudio',
    'normalizeAudio',
    'mirrorMpvVolume',
+    'reviewTiming',
  ] as const) {
    applyModernValue(
      context,
@@ -128,18 +129,28 @@ export function applyModernMediaResolution(
    'Expected non-negative number.',
  );

-  for (const key of ['fallbackDuration', 'maxMediaDuration'] as const) {
-    applyModernValue(
-      context,
-      media,
-      key,
-      `ankiConnect.media.${key}`,
-      asPositiveNumber,
-      DEFAULT_CONFIG.ankiConnect.media[key],
-      (value) => {
-        context.resolved.ankiConnect.media[key] = value;
-      },
-      'Expected positive number.',
-    );
-  }
+  applyModernValue(
+    context,
+    media,
+    'fallbackDuration',
+    'ankiConnect.media.fallbackDuration',
+    asPositiveNumber,
+    DEFAULT_CONFIG.ankiConnect.media.fallbackDuration,
+    (value) => {
+      context.resolved.ankiConnect.media.fallbackDuration = value;
+    },
+    'Expected positive number.',
+  );
+  applyModernValue(
+    context,
+    media,
+    'maxMediaDuration',
+    'ankiConnect.media.maxMediaDuration',
+    asNonNegativeNumber,
+    DEFAULT_CONFIG.ankiConnect.media.maxMediaDuration,
+    (value) => {
+      context.resolved.ankiConnect.media.maxMediaDuration = value;
+    },
+    'Expected non-negative number.',
+  );
 }
@@ -0,0 +1,34 @@
+import { DEFAULT_CONFIG } from '../../definitions';
+import type { ResolveContext } from '../context';
+
+export function applyAnkiSenrenResolution(context: ResolveContext): void {
+  if (
+    context.resolved.ankiConnect.isSenren.fieldGrouping !== 'auto' &&
+    context.resolved.ankiConnect.isSenren.fieldGrouping !== 'manual' &&
+    context.resolved.ankiConnect.isSenren.fieldGrouping !== 'disabled'
+  ) {
+    context.warn(
+      'ankiConnect.isSenren.fieldGrouping',
+      context.resolved.ankiConnect.isSenren.fieldGrouping,
+      DEFAULT_CONFIG.ankiConnect.isSenren.fieldGrouping,
+      'Expected auto, manual, or disabled.',
+    );
+    context.resolved.ankiConnect.isSenren.fieldGrouping =
+      DEFAULT_CONFIG.ankiConnect.isSenren.fieldGrouping;
+  }
+
+  // Kiku and Senren field grouping write incompatible markup into the same note
+  // fields, so only one may be active; Kiku wins to preserve pre-existing setups.
+  if (
+    context.resolved.ankiConnect.isSenren.enabled === true &&
+    context.resolved.ankiConnect.isKiku.enabled === true
+  ) {
+    context.warn(
+      'ankiConnect.isSenren.enabled',
+      true,
+      false,
+      'Kiku and Senren are mutually exclusive; disable isKiku.enabled to use Senren field grouping.',
+    );
+    context.resolved.ankiConnect.isSenren.enabled = false;
+  }
+}
@@ -298,10 +298,12 @@ test('settings registry puts feature toggles first, then other toggles alphabeti
  ];
  assert.equal(miningSections[0], 'AnkiConnect');

-  const kikuLapis = fields.filter((candidate) => candidate.section === 'Kiku/Lapis Features');
+  const kikuLapis = fields.filter(
+    (candidate) => candidate.section === 'Kiku/Lapis/Senren Features',
+  );
  assert.deepEqual(
-    kikuLapis.slice(0, 2).map((candidate) => candidate.configPath),
-    ['ankiConnect.isLapis.enabled', 'ankiConnect.isKiku.enabled'],
+    kikuLapis.slice(0, 3).map((candidate) => candidate.configPath),
+    ['ankiConnect.isLapis.enabled', 'ankiConnect.isKiku.enabled', 'ankiConnect.isSenren.enabled'],
  );
 });

@@ -352,6 +354,7 @@ test('settings registry marks safe live config paths as hot-reloadable', () => {
    'ankiConnect.deck',
    'ankiConnect.media.normalizeAudio',
    'ankiConnect.media.mirrorMpvVolume',
+    'ankiConnect.media.reviewTiming',
    'ankiConnect.knownWords.highlightEnabled',
    'ankiConnect.knownWords.refreshMinutes',
    'ankiConnect.knownWords.addMinedWordsImmediately',
@@ -366,6 +369,7 @@ test('settings registry marks safe live config paths as hot-reloadable', () => {
    'ankiConnect.fields.miscInfo',
    'ankiConnect.isLapis.sentenceCardModel',
    'ankiConnect.isKiku.fieldGrouping',
+    'ankiConnect.isSenren.fieldGrouping',
  ]) {
    assert.equal(field(path).restartBehavior, 'hot-reload', path);
  }
@@ -131,7 +131,7 @@ const SECTION_ORDER = new Map<string, number>(
    'AnkiConnect',
    'Note Fields',
    'Media Capture',
-    'Kiku/Lapis Features',
+    'Kiku/Lapis/Senren Features',
    'Anki AI',
    'AnkiConnect Proxy',
    'Jimaku',
@@ -163,6 +163,7 @@ const PATH_ORDER = new Map<string, number>(
    'ankiConnect.proxy.enabled',
    'ankiConnect.isLapis.enabled',
    'ankiConnect.isKiku.enabled',
+    'ankiConnect.isSenren.enabled',
    'subtitleStyle.knownWordColor',
    'ankiConnect.knownWords.matureThresholdDays',
    'subtitleStyle.knownWordMaturityColors.new',
@@ -221,6 +222,7 @@ const LABEL_OVERRIDES: Record<string, string> = {
  'ankiConnect.nPlusOne.enabled': 'Enabled',
  'ankiConnect.isLapis.enabled': 'Enable Lapis Features',
  'ankiConnect.isKiku.enabled': 'Enable Kiku Features',
+  'ankiConnect.isSenren.enabled': 'Enable Senren Features',
  'ankiConnect.lapisKiku.wordCardKind': 'Word Card Type',
  'stats.toggleKey': 'Toggle Stats Overlay',
  'shortcuts.openCharacterDictionaryManager': 'Open Character Dictionary Manager',
@@ -244,6 +246,7 @@ const LABEL_OVERRIDES: Record<string, string> = {
  'mpv.aniskipEnabled': 'Enable AniSkip',
  'mpv.aniskipButtonKey': 'AniSkip Button Key',
  'ankiConnect.media.mirrorMpvVolume': 'Mirror mpv Volume',
+  'ankiConnect.media.reviewTiming': 'Review Media Timing',
  'discordPresence.updateIntervalMs': 'Update Interval (ms)',
 };

@@ -251,7 +254,9 @@ const DESCRIPTION_OVERRIDES: Record<string, string> = {
  'ankiConnect.pollingRate':
    'Polling interval in milliseconds. Ignored while the local AnkiConnect proxy is enabled because push-based enrichment is used instead.',
  'ankiConnect.isKiku.enabled':
-    'Enable Kiku-specific mining behavior. Kiku supersedes Lapis: Lapis features still work, and Kiku adds duplicate handling and field grouping.',
+    'Enable Kiku-specific mining behavior. Kiku supersedes Lapis: Lapis features still work, and Kiku adds duplicate handling and field grouping. Mutually exclusive with Senren.',
+  'ankiConnect.isSenren.enabled':
+    'Enable Senren-specific duplicate handling: field grouping merges duplicates into Senren scene-switching markup (including miscInfo grouping). Mutually exclusive with Kiku; only one can be enabled at a time.',
  'ankiConnect.isLapis.enabled':
    'Enable Lapis-specific mining behavior and sentence-card model targeting. When Kiku is enabled, Lapis features still work and Kiku-specific features are added on top.',
  'ankiConnect.isLapis.sentenceCardModel':
@@ -407,9 +412,10 @@ function categoryAndSection(path: string): { category: ConfigSettingsCategory; s
  if (
    path.startsWith('ankiConnect.isKiku.') ||
    path.startsWith('ankiConnect.isLapis.') ||
+    path.startsWith('ankiConnect.isSenren.') ||
    path.startsWith('ankiConnect.lapisKiku.')
  ) {
-    return { category: 'mining-anki', section: 'Kiku/Lapis Features' };
+    return { category: 'mining-anki', section: 'Kiku/Lapis/Senren Features' };
  }
  if (path.startsWith('ankiConnect.ai.')) {
    return { category: 'mining-anki', section: 'Anki AI' };
@@ -694,6 +700,7 @@ function restartBehaviorForPath(path: string): ConfigSettingsRestartBehavior {
    path === 'ankiConnect.ai.enabled' ||
    path === 'ankiConnect.media.normalizeAudio' ||
    path === 'ankiConnect.media.mirrorMpvVolume' ||
+    path === 'ankiConnect.media.reviewTiming' ||
    path === 'ankiConnect.behavior.autoUpdateNewCards' ||
    path === 'ankiConnect.knownWords.highlightEnabled' ||
    path === 'ankiConnect.knownWords.refreshMinutes' ||
@@ -709,6 +716,7 @@ function restartBehaviorForPath(path: string): ConfigSettingsRestartBehavior {
    path === 'ankiConnect.fields.miscInfo' ||
    path === 'ankiConnect.isLapis.sentenceCardModel' ||
    path === 'ankiConnect.isKiku.fieldGrouping' ||
+    path === 'ankiConnect.isSenren.fieldGrouping' ||
    path === 'ankiConnect.lapisKiku.wordCardKind' ||
    path === 'mpv.aniskipEnabled' ||
    path === 'mpv.aniskipButtonKey' ||
@@ -65,6 +65,7 @@ export interface AnkiJimakuIpcRuntimeOptions {
  getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
  showDesktopNotification: (title: string, options: { body?: string; icon?: string }) => void;
  showOverlayNotification?: (payload: OverlayNotificationPayload) => void;
+  dismissOverlayNotification?: (id: string) => void;
  createFieldGroupingCallback: () => (
    data: KikuFieldGroupingRequestData,
  ) => Promise<KikuFieldGroupingChoice>;
@@ -166,6 +167,7 @@ export function registerAnkiJimakuIpcRuntime(
          options.getCachedMediaPath,
          options.shouldRequireRemoteMediaCache,
          options.getYoutubeMediaSourceUrl,
+          options.dismissOverlayNotification,
        );
        integration.start();
        options.setAnkiIntegration(integration);
@@ -10,6 +10,8 @@ import {
  isAssTemporalCommand,
  normalizePlainSubtitleText,
  parseAssEffectField,
+  removeLiveGlyphFragmentLines,
+  removeAssControlDebrisLines,
 } from './ass-text';

 test('assToPlainText drops vector drawing runs', () => {
@@ -74,6 +76,14 @@ test('assToPlainText normalizes CRLF before converting', () => {
  assert.equal(assToPlainText('一行目\r\n二行目'), '一行目\n二行目');
 });

+test('removeAssControlDebrisLines drops malformed spacer resets without eating dialogue', () => {
+  assert.equal(
+    removeAssControlDebrisLines('Visible line\n\\\n{\\fr0\n\\{\\frz287.5'),
+    'Visible line',
+  );
+  assert.equal(removeAssControlDebrisLines('本文{\\pos(1,2)'), '本文{\\pos(1,2)');
+});
+
 test('normalizePlainSubtitleText settles whitespace without decoding ASS', () => {
  // A brace reaching this layer is literal text mpv chose to show, not markup.
  assert.equal(normalizePlainSubtitleText('本文{\\pos(1,2)'), '本文{\\pos(1,2)');
@@ -193,3 +203,33 @@ test('isAnimatedAssEffectKind covers the stock animated effects only', () => {
  assert.equal(isAnimatedAssEffectKind('other'), false);
  assert.equal(isAnimatedAssEffectKind('none'), false);
 });
+
+test('removeLiveGlyphFragmentLines drops a per-glyph typesetting wall and its syllable', () => {
+  const wall = [...'wansdumretoikhI'].join('\n');
+  assert.equal(removeLiveGlyphFragmentLines(`${wall}\ntai`), '');
+});
+
+test('removeLiveGlyphFragmentLines keeps concurrent dialogue beside a glyph wall', () => {
+  const wall = [...'wansdumretoikhI'].join('\n');
+  assert.equal(removeLiveGlyphFragmentLines(`${wall}\nそれよりも　ノート…`), 'それよりも　ノート…');
+});
+
+test('removeLiveGlyphFragmentLines leaves ordinary short lines alone', () => {
+  const text = 'え\nはい。\nそうだな';
+  assert.equal(removeLiveGlyphFragmentLines(text), text);
+});
+
+test('normalizePlainSubtitleText folds cue-boundary blank lines for text consumers', () => {
+  // The display layer splits on the blank line before normalizing; everyone else --
+  // tokenizer, cache key, dedup gate, mined sentence -- wants the plain line form.
+  assert.equal(
+    normalizePlainSubtitleText('\u4e00\u884c\u76ee\n\n\u4e8c\u884c\u76ee'),
+    '\u4e00\u884c\u76ee\n\u4e8c\u884c\u76ee',
+  );
+  assert.equal(
+    normalizePlainSubtitleText('\u4e00\u884c\u76ee\n\n\u4e8c\u884c\u76ee', {
+      collapseLineBreaks: true,
+    }),
+    '\u4e00\u884c\u76ee \u4e8c\u884c\u76ee',
+  );
+});
@@ -91,6 +91,41 @@ export function assToPlainText(text: string, lineBreak: AssLineBreak = '\n'): st
  return resolveWhitespaceEscapes(stripAssMarkup(text.replace(/\r\n/g, '\n')), lineBreak);
 }

+const MALFORMED_ASS_ROTATION_RESET = /^\\?\{\\(?:fr|frx|fry|frz|fax|fay)[-+.0-9]*$/u;
+
+/**
+ * Drop non-rendering spacer events left as literal text by a malformed, unclosed ASS
+ * rotation reset. These events otherwise become repeated `\\` or `{\\fr0` subtitle
+ * lines after mpv-compatible decoding.
+ */
+export function removeAssControlDebrisLines(text: string): string {
+  return text
+    .split('\n')
+    .filter((line) => {
+      const compact = line.replace(/\s+/gu, '');
+      return compact !== '\\' && !MALFORMED_ASS_ROTATION_RESET.test(compact);
+    })
+    .join('\n');
+}
+
+const MIN_GLYPH_BURST_LINES = 6;
+const MAX_GLYPH_BURST_COMPANION_GLYPHS = 3;
+
+/**
+ * Per-glyph karaoke typesetting flattened into live text becomes a wall of
+ * single-character lines plus the short syllable currently being typed. No authored
+ * subtitle stacks this many one-glyph lines at once, so when the wall is present drop
+ * it and its short companion fragments while keeping any concurrent dialogue line.
+ */
+export function removeLiveGlyphFragmentLines(text: string): string {
+  const lines = text.split('\n');
+  const singleGlyphLines = lines.filter((line) => [...line.trim()].length === 1).length;
+  if (singleGlyphLines < MIN_GLYPH_BURST_LINES) return text;
+  return lines
+    .filter((line) => [...line.trim()].length > MAX_GLYPH_BURST_COMPANION_GLYPHS)
+    .join('\n');
+}
+
 export interface NormalizePlainSubtitleTextOptions {
  /** Fold every line break into a single space. */
  collapseLineBreaks?: boolean;
@@ -118,6 +153,10 @@ export function normalizePlainSubtitleText(
  );
  if (collapseLineBreaks) {
    normalized = normalized.replace(/\n/g, ' ').replace(/\s+/g, ' ');
+  } else {
+    // Simultaneous cues reach the display layer separated by a blank line; every other
+    // consumer wants the plain one-break-per-line form.
+    normalized = normalized.replace(/\n{2,}/g, '\n');
  }

  return trim ? normalized.trim() : normalized;
@@ -33,6 +33,7 @@ test('classifyConfigHotReloadDiff treats safe nested config paths as hot-reloada
  next.ankiConnect.deck = 'Mining';
  next.ankiConnect.media.normalizeAudio = !prev.ankiConnect.media.normalizeAudio;
  next.ankiConnect.media.mirrorMpvVolume = !prev.ankiConnect.media.mirrorMpvVolume;
+  next.ankiConnect.media.reviewTiming = !prev.ankiConnect.media.reviewTiming;
  next.ankiConnect.behavior.autoUpdateNewCards = !prev.ankiConnect.behavior.autoUpdateNewCards;
  next.ankiConnect.knownWords.highlightEnabled = !prev.ankiConnect.knownWords.highlightEnabled;
  next.ankiConnect.knownWords.refreshMinutes = prev.ankiConnect.knownWords.refreshMinutes + 5;
@@ -69,6 +70,7 @@ test('classifyConfigHotReloadDiff treats safe nested config paths as hot-reloada
      'ankiConnect.deck',
      'ankiConnect.media.normalizeAudio',
      'ankiConnect.media.mirrorMpvVolume',
+      'ankiConnect.media.reviewTiming',
      'ankiConnect.behavior.autoUpdateNewCards',
      'ankiConnect.knownWords.highlightEnabled',
      'ankiConnect.knownWords.refreshMinutes',
@@ -70,6 +70,7 @@ const HOT_RELOAD_EXACT_OR_PREFIX_PATHS = [
  'ankiConnect.deck',
  'ankiConnect.media.normalizeAudio',
  'ankiConnect.media.mirrorMpvVolume',
+  'ankiConnect.media.reviewTiming',
  'ankiConnect.behavior.autoUpdateNewCards',
  'ankiConnect.knownWords.highlightEnabled',
  'ankiConnect.knownWords.refreshMinutes',
@@ -85,6 +86,7 @@ const HOT_RELOAD_EXACT_OR_PREFIX_PATHS = [
  'ankiConnect.fields.miscInfo',
  'ankiConnect.isLapis.sentenceCardModel',
  'ankiConnect.isKiku.fieldGrouping',
+  'ankiConnect.isSenren.fieldGrouping',
  'ankiConnect.lapisKiku.wordCardKind',
 ] as const;

@@ -306,9 +306,11 @@ test('vocabulary charts use complete top-word and lexical rollup data', () => {
      `INSERT INTO imm_words(headword, word, reading, first_seen, last_seen, frequency)
       VALUES (?, ?, '', 1700000000, 1700000000, ?)`,
    );
+    db.exec('BEGIN');
    for (let index = 0; index < 501; index += 1) {
      insertWord.run(`語${index}`, `語${index}`, index === 500 ? 10_000 : 1);
    }
+    db.exec('COMMIT');

    const charts = getVocabularyChartData(db);

@@ -3,7 +3,7 @@ import fs from 'node:fs';
 import os from 'node:os';
 import path from 'node:path';
 import test from 'node:test';
-import { Database } from './sqlite';
+import { Database, type DatabaseSync } from './sqlite';
 import { getStatsExcludedWords, replaceStatsExcludedWords } from './query-lexical';
 import { finalizeSessionRecord, startSessionRecord } from './session';
 import {
@@ -87,6 +87,29 @@ test('applyPragmas sets the SQLite tuning defaults used by immersion tracking',
  }
 });

+test('applyPragmas installs the busy timeout before WAL negotiation', () => {
+  const statements: string[] = [];
+  const db: DatabaseSync = {
+    exec(source) {
+      statements.push(source);
+      return db;
+    },
+    prepare() {
+      throw new Error('not used');
+    },
+    close() {
+      return db;
+    },
+  };
+
+  applyPragmas(db);
+
+  assert.deepEqual(statements.slice(0, 2), [
+    'PRAGMA busy_timeout = 2500',
+    'PRAGMA journal_mode = WAL',
+  ]);
+});
+
 test('ensureSchema creates immersion core tables', () => {
  const dbPath = makeDbPath();
  const db = new Database(dbPath);
@@ -315,10 +315,12 @@ function migrateSessionEventTimestampsToText(db: DatabaseSync): void {
 }

 export function applyPragmas(db: DatabaseSync): void {
+  // Install the wait policy before WAL negotiation, which can briefly contend with
+  // another connection closing or checkpointing the same database.
+  db.exec('PRAGMA busy_timeout = 2500');
  db.exec('PRAGMA journal_mode = WAL');
  db.exec('PRAGMA synchronous = NORMAL');
  db.exec('PRAGMA foreign_keys = ON');
-  db.exec('PRAGMA busy_timeout = 2500');
  db.exec(`PRAGMA journal_size_limit = ${WAL_JOURNAL_SIZE_LIMIT_BYTES}`);
 }

@@ -50,6 +50,11 @@ export {
 } from './tokenizer/yomitan-parser-runtime';
 export { syncYomitanDefaultAnkiServer } from './tokenizer/yomitan-parser-runtime';
 export { createSubtitleProcessingController } from './subtitle-processing-controller';
+export {
+  resolveSanitizedSubtitleSeekCommand,
+  subtitleCueListSeekTime,
+  subtitleCueSeekTime,
+} from './subtitle-cue-navigation';
 export { createFrequencyDictionaryLookup } from './frequency-dictionary';
 export { createJlptVocabularyLookup } from './jlpt-vocab';
 export {
@@ -126,12 +131,6 @@ export {
  resolvePlaybackPlan as resolveJellyfinPlaybackPlanRuntime,
  ticksToSeconds as jellyfinTicksToSecondsRuntime,
 } from './jellyfin';
-export { loadJellyfinSubtitleDelay, saveJellyfinSubtitleDelay } from './jellyfin-subtitle-delay';
-export {
-  estimateSubtitleTimingOffset,
-  type SubtitleTimingOffsetOptions,
-  type SubtitleTimingOffsetResult,
-} from './subtitle-timing-offset';
 export { buildJellyfinTimelinePayload, JellyfinRemoteSessionService } from './jellyfin-remote';
 export {
  broadcastRuntimeOptionsChangedRuntime,
@@ -648,6 +648,83 @@ test('registerIpcHandlers exposes playback window activation request', async ()
  assert.deepEqual(calls, ['activate']);
 });

+test('registerIpcHandlers accepts the keep-without-media timing decision', async () => {
+  const { registrar, handlers } = createFakeIpcRegistrar();
+  const requests: unknown[] = [];
+  registerIpcHandlers(
+    createRegisterIpcDeps({
+      resolveMediaTimingReview: async (request) => {
+        requests.push(request);
+        return { ok: true };
+      },
+    }),
+    registrar,
+  );
+
+  const handler = handlers.handle.get(IPC_CHANNELS.request.mediaTimingReviewResolve);
+  assert.ok(handler);
+  assert.deepEqual(
+    await handler!({}, { reviewId: 'review-1', decision: { action: 'skip-media' } }),
+    { ok: true },
+  );
+  assert.deepEqual(requests, [{ reviewId: 'review-1', decision: { action: 'skip-media' } }]);
+});
+
+test('registerIpcHandlers validates and forwards combined timing review text', async () => {
+  const { registrar, handlers } = createFakeIpcRegistrar();
+  const requests: unknown[] = [];
+  registerIpcHandlers(
+    createRegisterIpcDeps({
+      resolveMediaTimingReview: async (request) => {
+        requests.push(request);
+        return { ok: true };
+      },
+    }),
+    registrar,
+  );
+
+  const handler = handlers.handle.get(IPC_CHANNELS.request.mediaTimingReviewResolve);
+  assert.ok(handler);
+  assert.deepEqual(
+    await handler!(
+      {},
+      {
+        reviewId: 'review-1',
+        decision: {
+          action: 'confirm',
+          startTime: 10,
+          endTime: 12,
+          text: '前の行 対象の行',
+        },
+      },
+    ),
+    { ok: true },
+  );
+  assert.deepEqual(requests, [
+    {
+      reviewId: 'review-1',
+      decision: {
+        action: 'confirm',
+        startTime: 10,
+        endTime: 12,
+        text: '前の行 対象の行',
+      },
+    },
+  ]);
+
+  assert.deepEqual(
+    await handler!(
+      {},
+      {
+        reviewId: 'review-1',
+        decision: { action: 'confirm', startTime: 10, endTime: 12, text: '   ' },
+      },
+    ),
+    { ok: false, message: 'Timing review is unavailable.' },
+  );
+  assert.equal(requests.length, 1);
+});
+
 test('registerIpcHandlers forwards yomitan lookup tracking commands to immersion tracker', () => {
  const { registrar, handlers } = createFakeIpcRegistrar();
  const calls: string[] = [];
@@ -19,6 +19,13 @@ import type {
  YoutubePickerResolveRequest,
  YoutubePickerResolveResult,
 } from '../../types';
+import type {
+  MediaTimingReviewActionResult,
+  MediaTimingReviewPreviewRequest,
+  MediaTimingReviewResolveRequest,
+  MediaTimingReviewWaveformRequest,
+  MediaTimingReviewWaveformResult,
+} from '../../types/anki';
 import { IPC_CHANNELS, type OverlayHostedModal } from '../../shared/ipc/contracts';
 import {
  parseMpvCommand,
@@ -99,6 +106,16 @@ export interface IpcServiceDeps {
  onYoutubePickerResolve: (
    request: YoutubePickerResolveRequest,
  ) => Promise<YoutubePickerResolveResult>;
+  previewMediaTimingReview?: (
+    request: MediaTimingReviewPreviewRequest,
+  ) => Promise<MediaTimingReviewActionResult>;
+  getMediaTimingReviewWaveform?: (
+    request: MediaTimingReviewWaveformRequest,
+  ) => Promise<MediaTimingReviewWaveformResult>;
+  stopMediaTimingReviewPreview?: (reviewId: string) => Promise<MediaTimingReviewActionResult>;
+  resolveMediaTimingReview?: (
+    request: MediaTimingReviewResolveRequest,
+  ) => MediaTimingReviewActionResult | Promise<MediaTimingReviewActionResult>;
  getAnkiConnectStatus: () => boolean;
  getRuntimeOptions: () => unknown;
  setRuntimeOption: (id: RuntimeOptionId, value: RuntimeOptionValue) => unknown;
@@ -222,6 +239,72 @@ function parseOverlayNotificationActionPayload(
  return { notificationId, actionId, ...(typeof noteId === 'number' ? { noteId } : {}) };
 }

+function parseMediaTimingReviewPreviewRequest(
+  payload: unknown,
+): MediaTimingReviewPreviewRequest | null {
+  if (!payload || typeof payload !== 'object') return null;
+  const record = payload as Record<string, unknown>;
+  if (
+    typeof record.reviewId !== 'string' ||
+    !record.reviewId ||
+    typeof record.startTime !== 'number' ||
+    !Number.isFinite(record.startTime) ||
+    typeof record.endTime !== 'number' ||
+    !Number.isFinite(record.endTime)
+  ) {
+    return null;
+  }
+  return {
+    reviewId: record.reviewId,
+    startTime: record.startTime,
+    endTime: record.endTime,
+  };
+}
+
+function parseMediaTimingReviewWaveformRequest(
+  payload: unknown,
+): MediaTimingReviewWaveformRequest | null {
+  return parseMediaTimingReviewPreviewRequest(payload);
+}
+
+function parseMediaTimingReviewResolveRequest(
+  payload: unknown,
+): MediaTimingReviewResolveRequest | null {
+  if (!payload || typeof payload !== 'object') return null;
+  const record = payload as Record<string, unknown>;
+  if (typeof record.reviewId !== 'string' || !record.reviewId) return null;
+  const decision = record.decision;
+  if (!decision || typeof decision !== 'object') return null;
+  const decisionRecord = decision as Record<string, unknown>;
+  if (
+    decisionRecord.action === 'use-original' ||
+    decisionRecord.action === 'skip-media' ||
+    decisionRecord.action === 'discard'
+  ) {
+    return { reviewId: record.reviewId, decision: { action: decisionRecord.action } };
+  }
+  if (
+    decisionRecord.action === 'confirm' &&
+    typeof decisionRecord.startTime === 'number' &&
+    Number.isFinite(decisionRecord.startTime) &&
+    typeof decisionRecord.endTime === 'number' &&
+    Number.isFinite(decisionRecord.endTime) &&
+    (decisionRecord.text === undefined ||
+      (typeof decisionRecord.text === 'string' && decisionRecord.text.trim().length > 0))
+  ) {
+    return {
+      reviewId: record.reviewId,
+      decision: {
+        action: 'confirm',
+        startTime: decisionRecord.startTime,
+        endTime: decisionRecord.endTime,
+        ...(decisionRecord.text === undefined ? {} : { text: decisionRecord.text }),
+      },
+    };
+  }
+  return null;
+}
+
 export interface IpcDepsRuntimeOptions {
  getMainWindow: () => WindowLike | null;
  getVisibleOverlayVisibility: () => boolean;
@@ -278,6 +361,10 @@ export interface IpcDepsRuntimeOptions {
  onYoutubePickerResolve: (
    request: YoutubePickerResolveRequest,
  ) => Promise<YoutubePickerResolveResult>;
+  previewMediaTimingReview?: IpcServiceDeps['previewMediaTimingReview'];
+  getMediaTimingReviewWaveform?: IpcServiceDeps['getMediaTimingReviewWaveform'];
+  stopMediaTimingReviewPreview?: IpcServiceDeps['stopMediaTimingReviewPreview'];
+  resolveMediaTimingReview?: IpcServiceDeps['resolveMediaTimingReview'];
  getAnkiConnectStatus: () => boolean;
  getRuntimeOptions: () => unknown;
  setRuntimeOption: (id: RuntimeOptionId, value: RuntimeOptionValue) => unknown;
@@ -371,6 +458,10 @@ export function createIpcDepsRuntime(options: IpcDepsRuntimeOptions): IpcService
      options.activatePlaybackWindowForOverlayInteraction ?? (() => false),
    runSubsyncManual: options.runSubsyncManual,
    onYoutubePickerResolve: options.onYoutubePickerResolve,
+    previewMediaTimingReview: options.previewMediaTimingReview,
+    getMediaTimingReviewWaveform: options.getMediaTimingReviewWaveform,
+    stopMediaTimingReviewPreview: options.stopMediaTimingReviewPreview,
+    resolveMediaTimingReview: options.resolveMediaTimingReview,
    getAnkiConnectStatus: options.getAnkiConnectStatus,
    getRuntimeOptions: options.getRuntimeOptions,
    setRuntimeOption: options.setRuntimeOption,
@@ -498,6 +589,46 @@ export function registerIpcHandlers(deps: IpcServiceDeps, ipc: IpcMainRegistrar
    },
  );

+  ipc.handle(
+    IPC_CHANNELS.request.mediaTimingReviewPreview,
+    async (_event: unknown, payload: unknown) => {
+      const request = parseMediaTimingReviewPreviewRequest(payload);
+      if (!request || !deps.previewMediaTimingReview) {
+        return { ok: false, message: 'Timing preview is unavailable.' };
+      }
+      return await deps.previewMediaTimingReview(request);
+    },
+  );
+  ipc.handle(
+    IPC_CHANNELS.request.mediaTimingReviewWaveform,
+    async (_event: unknown, payload: unknown) => {
+      const request = parseMediaTimingReviewWaveformRequest(payload);
+      if (!request || !deps.getMediaTimingReviewWaveform) {
+        return { ok: false, message: 'Timing waveform is unavailable.' };
+      }
+      return await deps.getMediaTimingReviewWaveform(request);
+    },
+  );
+  ipc.handle(
+    IPC_CHANNELS.request.mediaTimingReviewStopPreview,
+    async (_event: unknown, reviewId: unknown) => {
+      if (typeof reviewId !== 'string' || !reviewId || !deps.stopMediaTimingReviewPreview) {
+        return { ok: false, message: 'Timing preview is unavailable.' };
+      }
+      return await deps.stopMediaTimingReviewPreview(reviewId);
+    },
+  );
+  ipc.handle(
+    IPC_CHANNELS.request.mediaTimingReviewResolve,
+    async (_event: unknown, payload: unknown) => {
+      const request = parseMediaTimingReviewResolveRequest(payload);
+      if (!request || !deps.resolveMediaTimingReview) {
+        return { ok: false, message: 'Timing review is unavailable.' };
+      }
+      return await deps.resolveMediaTimingReview(request);
+    },
+  );
+
  ipc.on(IPC_CHANNELS.command.openYomitanSettings, () => {
    deps.openYomitanSettings();
  });
@@ -1,54 +0,0 @@
-import assert from 'node:assert/strict';
-import * as fs from 'node:fs';
-import * as os from 'node:os';
-import * as path from 'node:path';
-import test from 'node:test';
-import { loadJellyfinSubtitleDelay, saveJellyfinSubtitleDelay } from './jellyfin-subtitle-delay';
-
-function statePath(name: string): string {
-  return path.join(fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-jellyfin-delay-')), name);
-}
-
-test('jellyfin subtitle delay store saves and loads delay by item and stream', () => {
-  const filePath = statePath('delays.json');
-
-  assert.equal(
-    saveJellyfinSubtitleDelay({
-      filePath,
-      itemId: 'episode-1',
-      streamIndex: 3,
-      delaySeconds: 1.25,
-    }),
-    true,
-  );
-
-  assert.equal(loadJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 3 }), 1.25);
-  assert.equal(loadJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 4 }), null);
-});
-
-test('jellyfin subtitle delay store preserves other stream delays when updating one stream', () => {
-  const filePath = statePath('delays.json');
-
-  saveJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 3, delaySeconds: 1.25 });
-  saveJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 4, delaySeconds: -0.5 });
-  saveJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 3, delaySeconds: 2 });
-
-  assert.equal(loadJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 3 }), 2);
-  assert.equal(loadJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 4 }), -0.5);
-});
-
-test('jellyfin subtitle delay store ignores invalid files and values', () => {
-  const filePath = statePath('delays.json');
-  fs.writeFileSync(filePath, '{');
-
-  assert.equal(loadJellyfinSubtitleDelay({ filePath, itemId: 'episode-1', streamIndex: 3 }), null);
-  assert.equal(
-    saveJellyfinSubtitleDelay({
-      filePath,
-      itemId: 'episode-1',
-      streamIndex: 3,
-      delaySeconds: Number.NaN,
-    }),
-    false,
-  );
-});
@@ -1,66 +0,0 @@
-import * as fs from 'fs';
-import * as path from 'path';
-
-type JellyfinSubtitleDelayStore = {
-  version?: unknown;
-  delays?: unknown;
-};
-
-type JellyfinSubtitleDelayParams = {
-  filePath: string;
-  itemId: string;
-  streamIndex: number;
-};
-
-type SaveJellyfinSubtitleDelayParams = JellyfinSubtitleDelayParams & {
-  delaySeconds: number;
-};
-
-function storeKey(itemId: string, streamIndex: number): string {
-  return JSON.stringify([itemId, streamIndex]);
-}
-
-function readDelayMap(filePath: string): Record<string, number> {
-  try {
-    if (!fs.existsSync(filePath)) return {};
-    const parsed = JSON.parse(fs.readFileSync(filePath, 'utf-8')) as JellyfinSubtitleDelayStore;
-    if (
-      !parsed ||
-      typeof parsed !== 'object' ||
-      !parsed.delays ||
-      typeof parsed.delays !== 'object'
-    ) {
-      return {};
-    }
-    const delays: Record<string, number> = {};
-    for (const [key, value] of Object.entries(parsed.delays as Record<string, unknown>)) {
-      if (typeof value === 'number' && Number.isFinite(value)) {
-        delays[key] = value;
-      }
-    }
-    return delays;
-  } catch {
-    return {};
-  }
-}
-
-export function loadJellyfinSubtitleDelay(params: JellyfinSubtitleDelayParams): number | null {
-  const delay = readDelayMap(params.filePath)[storeKey(params.itemId, params.streamIndex)];
-  return typeof delay === 'number' && Number.isFinite(delay) ? delay : null;
-}
-
-export function saveJellyfinSubtitleDelay(params: SaveJellyfinSubtitleDelayParams): boolean {
-  if (!Number.isFinite(params.delaySeconds)) return false;
-  try {
-    const delays = readDelayMap(params.filePath);
-    delays[storeKey(params.itemId, params.streamIndex)] = params.delaySeconds;
-    const dir = path.dirname(params.filePath);
-    if (!fs.existsSync(dir)) {
-      fs.mkdirSync(dir, { recursive: true });
-    }
-    fs.writeFileSync(params.filePath, JSON.stringify({ version: 1, delays }, null, 2));
-    return true;
-  } catch {
-    return false;
-  }
-}
@@ -0,0 +1,291 @@
+import assert from 'node:assert/strict';
+import { EventEmitter } from 'node:events';
+import net from 'node:net';
+import { describe, test } from 'node:test';
+import { buildMediaTimingPreviewArgs, MediaTimingPreviewSession } from './media-timing-preview';
+
+describe('buildMediaTimingPreviewArgs', () => {
+  test('creates a hidden audio-only reusable mpv session', () => {
+    const args = buildMediaTimingPreviewArgs('/tmp/review.sock', {
+      mediaPath: '/video/show.mkv',
+      audioTrackId: 3,
+      volume: 55,
+    });
+
+    assert.ok(args.includes('--no-video'));
+    assert.ok(args.includes('--force-window=no'));
+    assert.ok(args.includes('--idle=yes'));
+    assert.ok(args.includes('--pause=yes'));
+    assert.ok(args.includes('--input-ipc-server=/tmp/review.sock'));
+    assert.ok(args.includes('--aid=3'));
+    assert.ok(args.includes('--volume=55'));
+    assert.equal(args.at(-2), '--');
+    assert.equal(args.at(-1), '/video/show.mkv');
+  });
+
+  test('keeps source timestamps for cached remote windows', () => {
+    const args = buildMediaTimingPreviewArgs('/tmp/review.sock', {
+      mediaPath: '/tmp/window.mkv',
+      absoluteTimestamps: true,
+    });
+
+    assert.ok(args.includes('--rebase-start-time=no'));
+    assert.equal(
+      buildMediaTimingPreviewArgs('/tmp/review.sock', { mediaPath: '/video/show.mkv' }).includes(
+        '--rebase-start-time=no',
+      ),
+      false,
+    );
+  });
+
+  test('separates an option-like media path without adding optional audio arguments', () => {
+    const args = buildMediaTimingPreviewArgs('/tmp/review.sock', {
+      mediaPath: '--fullscreen',
+    });
+
+    assert.equal(args.at(-2), '--');
+    assert.equal(args.at(-1), '--fullscreen');
+    assert.equal(
+      args.some((arg) => arg.startsWith('--aid=')),
+      false,
+    );
+    assert.equal(
+      args.some((arg) => arg.startsWith('--volume=')),
+      false,
+    );
+  });
+});
+
+test('preview session handles socket errors after connecting', async () => {
+  const socket = new net.Socket();
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => child as never,
+    connectSocket: () => {
+      queueMicrotask(() => socket.emit('connect'));
+      return socket;
+    },
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  await session.start({ mediaPath: '/video/show.mkv' });
+  assert.doesNotThrow(() => socket.emit('error', new Error('pipe closed')));
+  await assert.rejects(session.play(1, 2), /not ready/);
+  session.dispose();
+});
+
+test('preview session keeps failed connection errors handled through destruction', async () => {
+  const socket = new EventEmitter() as EventEmitter & {
+    destroy: () => void;
+  };
+  socket.destroy = () => {
+    socket.emit('error', new Error('socket failed again while closing'));
+  };
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  const times = [0, 0, 0, 6_000];
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => child as never,
+    connectSocket: () => {
+      queueMicrotask(() => socket.emit('error', new Error('connection failed')));
+      return socket as never;
+    },
+    now: () => times.shift() ?? 6_000,
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  await assert.rejects(session.start({ mediaPath: '/video/show.mkv' }), /Timed out starting/);
+});
+
+test('preview session rejects a connection that finishes after disposal', async () => {
+  const socket = new net.Socket();
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => child as never,
+    connectSocket: () => socket,
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  const pendingStart = session.start({ mediaPath: '-playlist' });
+  session.dispose();
+  socket.emit('connect');
+
+  await assert.rejects(pendingStart, /closed/);
+  assert.equal(socket.destroyed, true);
+});
+
+test('preview session shares one startup across concurrent start calls', async () => {
+  const socket = new net.Socket();
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  let spawnCount = 0;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => {
+      spawnCount += 1;
+      return child as never;
+    },
+    connectSocket: () => socket,
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  const firstStart = session.start({ mediaPath: '/video/show.mkv' });
+  const secondStart = session.start({ mediaPath: '/video/show.mkv' });
+  socket.emit('connect');
+
+  await Promise.all([firstStart, secondStart]);
+  assert.equal(spawnCount, 1);
+  session.dispose();
+});
+
+test('preview session can start again after a startup failure', async () => {
+  const socket = new net.Socket();
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  let spawnCount = 0;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => {
+      spawnCount += 1;
+      if (spawnCount === 1) throw new Error('spawn failed');
+      return child as never;
+    },
+    connectSocket: () => {
+      queueMicrotask(() => socket.emit('connect'));
+      return socket;
+    },
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  await assert.rejects(session.start({ mediaPath: '/video/show.mkv' }), /spawn failed/);
+  await session.start({ mediaPath: '/video/show.mkv' });
+  assert.equal(spawnCount, 2);
+  session.dispose();
+});
+
+test('preview session bounds a connection attempt that never settles', async () => {
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  let nowMs = 0;
+  let connectAttempts = 0;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => child as never,
+    connectSocket: () => {
+      connectAttempts += 1;
+      const socket = new net.Socket();
+      socket.destroy = (() => {
+        socket.emit('error', new Error('socket failed while timing out'));
+        return socket;
+      }) as typeof socket.destroy;
+      return socket;
+    },
+    now: () => {
+      const current = nowMs;
+      nowMs += 1_000;
+      return current;
+    },
+    schedule: (callback) => setTimeout(callback, 0),
+    cancelSchedule: (timeout) => clearTimeout(timeout),
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+
+  await assert.rejects(session.start({ mediaPath: '/video/show.mkv' }), /Timed out starting/);
+  assert.equal(connectAttempts, 1);
+});
+
+function createFakeSocket() {
+  const socket = new EventEmitter() as EventEmitter & {
+    destroyed: boolean;
+    write: (data: string) => boolean;
+    end: () => void;
+    destroy: () => void;
+    off: EventEmitter['off'];
+  };
+  const writes: string[] = [];
+  socket.destroyed = false;
+  socket.write = (data) => {
+    writes.push(data);
+    return true;
+  };
+  socket.end = () => undefined;
+  socket.destroy = () => {
+    socket.destroyed = true;
+  };
+  return { socket, writes };
+}
+
+test('preview session plays once to the clip end and reports when mpv has drained it', async () => {
+  const { socket, writes } = createFakeSocket();
+  const child = new EventEmitter() as EventEmitter & { kill: () => boolean };
+  child.kill = () => true;
+  const session = new MediaTimingPreviewSession({
+    platform: 'linux',
+    spawnProcess: () => child as never,
+    connectSocket: () => {
+      queueMicrotask(() => socket.emit('connect'));
+      return socket as never;
+    },
+    removeSocketFile: () => undefined,
+    createSocketPath: () => '/tmp/review.sock',
+  });
+  let endedCount = 0;
+  session.onPlaybackEnded(() => {
+    endedCount += 1;
+  });
+  const property = (name: string, data: boolean): string =>
+    `${JSON.stringify({ event: 'property-change', name, data })}\n`;
+
+  await session.start({ mediaPath: '/video/show.mkv' });
+  assert.deepEqual(
+    writes.map((line) => JSON.parse(line).command),
+    [
+      ['observe_property', 1, 'eof-reached'],
+      ['observe_property', 2, 'pause'],
+    ],
+  );
+  // The observers' initial replies describe the idle paused player, not a finished preview.
+  socket.emit('data', property('eof-reached', false) + property('pause', true));
+  assert.equal(endedCount, 0);
+
+  writes.length = 0;
+  await session.play(12.25, 14.5);
+  assert.deepEqual(
+    writes.map((line) => JSON.parse(line).command),
+    [
+      ['set_property', 'pause', true],
+      ['seek', 12.25, 'absolute+exact'],
+      ['set_property', 'end', '14.500'],
+      ['set_property', 'pause', false],
+    ],
+  );
+
+  // Events may arrive split across chunks. The decoder passing `end` flips eof-reached while
+  // audio still drains; only the keep-open pause that follows marks the preview as finished.
+  socket.emit('data', property('eof-reached', false) + property('pause', false).slice(0, 20));
+  socket.emit('data', property('pause', false).slice(20) + property('eof-reached', true));
+  assert.equal(endedCount, 0);
+  socket.emit('data', property('pause', true));
+  assert.equal(endedCount, 1);
+  socket.emit('data', property('pause', true));
+  assert.equal(endedCount, 1);
+
+  // Stopping early pauses without an end signal, and a later real EOF is not a preview end.
+  await session.play(1, 2);
+  socket.emit('data', property('eof-reached', false) + property('pause', false));
+  await session.stop();
+  socket.emit('data', property('pause', true) + property('eof-reached', true));
+  assert.equal(endedCount, 1);
+  session.dispose();
+});
@@ -0,0 +1,394 @@
+import { spawn, type ChildProcess } from 'child_process';
+import fs from 'fs';
+import net, { type Socket } from 'net';
+import os from 'os';
+import path from 'path';
+import { randomUUID } from 'crypto';
+
+const CONNECT_TIMEOUT_MS = 5_000;
+const CONNECT_ATTEMPT_TIMEOUT_MS = 500;
+const CONNECT_RETRY_MS = 40;
+/**
+ * mpv flips eof-reached as soon as the decoder passes `end`, while its audio buffer is still
+ * draining; keep-open then pauses once the buffer has played out. A preview has ended when
+ * both have happened.
+ */
+const EOF_OBSERVER_ID = 1;
+const PAUSE_OBSERVER_ID = 2;
+
+export interface MediaTimingPreviewStartOptions {
+  mediaPath: string;
+  executablePath?: string;
+  audioTrackId?: number;
+  volume?: number;
+  /** The file keeps source timestamps (a cached remote window); seek with the original times. */
+  absoluteTimestamps?: boolean;
+}
+
+type PreviewProcess = Pick<ChildProcess, 'kill' | 'once'>;
+
+interface MediaTimingPreviewDeps {
+  platform: NodeJS.Platform;
+  spawnProcess: (command: string, args: string[]) => PreviewProcess;
+  connectSocket: (socketPath: string) => Socket;
+  now: () => number;
+  schedule: (callback: () => void, delayMs: number) => ReturnType<typeof setTimeout>;
+  cancelSchedule: (timeout: ReturnType<typeof setTimeout>) => void;
+  removeSocketFile: (socketPath: string) => void;
+  createSocketPath: () => string;
+}
+
+export function buildMediaTimingPreviewArgs(
+  socketPath: string,
+  options: MediaTimingPreviewStartOptions,
+): string[] {
+  const args = [
+    '--no-config',
+    '--no-video',
+    '--audio-display=no',
+    '--force-window=no',
+    '--idle=yes',
+    '--keep-open=yes',
+    '--pause=yes',
+    '--terminal=no',
+    '--msg-level=all=warn',
+    `--input-ipc-server=${socketPath}`,
+  ];
+  if (typeof options.audioTrackId === 'number' && Number.isInteger(options.audioTrackId)) {
+    args.push(`--aid=${options.audioTrackId}`);
+  }
+  if (typeof options.volume === 'number' && Number.isFinite(options.volume)) {
+    args.push(`--volume=${Math.max(0, options.volume)}`);
+  }
+  if (options.absoluteTimestamps) {
+    args.push('--rebase-start-time=no');
+  }
+  args.push('--', options.mediaPath);
+  return args;
+}
+
+function createDefaultSocketPath(): string {
+  const suffix = `${process.pid}-${randomUUID()}`;
+  return process.platform === 'win32'
+    ? `\\\\.\\pipe\\subminer-timing-preview-${suffix}`
+    : path.join(
+        // macOS limits Unix socket paths to 104 bytes, while its temp directory can be long.
+        process.platform === 'darwin' ? '/tmp' : os.tmpdir(),
+        `subminer-timing-preview-${suffix}.sock`,
+      );
+}
+
+function removePosixSocketFile(socketPath: string): void {
+  if (process.platform === 'win32') return;
+  try {
+    fs.unlinkSync(socketPath);
+  } catch (error) {
+    if ((error as NodeJS.ErrnoException).code !== 'ENOENT') {
+      throw error;
+    }
+  }
+}
+
+export class MediaTimingPreviewSession {
+  private readonly deps: MediaTimingPreviewDeps;
+  private socketPath: string | null = null;
+  private socket: Socket | null = null;
+  private process: PreviewProcess | null = null;
+  private startupError: Error | null = null;
+  private startPromise: Promise<void> | null = null;
+  private retryWait: {
+    timeout: ReturnType<typeof setTimeout>;
+    resolve: () => void;
+  } | null = null;
+  private disposed = false;
+  private readBuffer = '';
+  private playing = false;
+  private eofReached = false;
+  private paused = true;
+  private readonly endedListeners = new Set<() => void>();
+
+  constructor(deps: Partial<MediaTimingPreviewDeps> = {}) {
+    this.deps = {
+      platform: process.platform,
+      spawnProcess: (command, args) => spawn(command, args, { stdio: 'ignore' }),
+      connectSocket: (socketPath) => net.createConnection(socketPath),
+      now: Date.now,
+      schedule: (callback, delayMs) => setTimeout(callback, delayMs),
+      cancelSchedule: (timeout) => clearTimeout(timeout),
+      removeSocketFile: removePosixSocketFile,
+      createSocketPath: createDefaultSocketPath,
+      ...deps,
+    };
+  }
+
+  async start(options: MediaTimingPreviewStartOptions): Promise<void> {
+    if (this.disposed) throw new Error('Preview session is closed');
+    if (this.socket) return;
+    if (this.startPromise) return await this.startPromise;
+
+    const startPromise = this.startOnce(options);
+    this.startPromise = startPromise;
+    try {
+      await startPromise;
+    } catch (error) {
+      this.releaseResources();
+      throw error;
+    } finally {
+      if (this.startPromise === startPromise) this.startPromise = null;
+    }
+  }
+
+  private async startOnce(options: MediaTimingPreviewStartOptions): Promise<void> {
+    const mediaPath = options.mediaPath.trim();
+    if (!mediaPath) throw new Error('No media source is available for preview');
+
+    const socketPath = this.deps.createSocketPath();
+    this.socketPath = socketPath;
+    if (this.deps.platform !== 'win32') {
+      this.deps.removeSocketFile(socketPath);
+    }
+
+    const command = options.executablePath?.trim() || 'mpv';
+    this.startupError = null;
+    const child = this.deps.spawnProcess(
+      command,
+      buildMediaTimingPreviewArgs(socketPath, { ...options, mediaPath }),
+    );
+    this.process = child;
+    child.once('error', (error) => {
+      if (this.process !== child) return;
+      this.startupError = error;
+    });
+    child.once('exit', () => {
+      if (this.process !== child) return;
+      if (!this.socket && !this.disposed && !this.startupError) {
+        this.startupError = new Error('The hidden mpv preview player exited during startup');
+      }
+      this.socket?.destroy();
+      this.socket = null;
+      this.process = null;
+    });
+
+    await this.connectWithRetry(socketPath);
+  }
+
+  /**
+   * Plays [startTime, endTime) once. mpv stops itself at `end` and, thanks to keep-open,
+   * pauses after draining the audio device, so the listener hears the whole clip even on
+   * high-latency outputs. onPlaybackEnded fires when mpv reports the end was reached.
+   */
+  async play(startTime: number, endTime: number): Promise<void> {
+    if (!this.socket || this.socket.destroyed) {
+      throw new Error('Preview player is not ready');
+    }
+    if (!Number.isFinite(startTime) || !Number.isFinite(endTime) || endTime <= startTime) {
+      throw new Error('Preview timing is invalid');
+    }
+
+    this.playing = false;
+    this.send(['set_property', 'pause', true]);
+    this.send(['seek', startTime, 'absolute+exact']);
+    // The option parser wants a time string; a raw JSON number is not accepted for `end`.
+    this.send(['set_property', 'end', endTime.toFixed(3)]);
+    this.send(['set_property', 'pause', false]);
+    // Only the seek's eof-reached=false and the later keep-open pause count for this play.
+    this.eofReached = false;
+    this.paused = false;
+    this.playing = true;
+  }
+
+  async stop(): Promise<void> {
+    this.playing = false;
+    if (!this.socket || this.socket.destroyed) return;
+    this.send(['set_property', 'pause', true]);
+  }
+
+  onPlaybackEnded(listener: () => void): void {
+    this.endedListeners.add(listener);
+  }
+
+  private finishPlayback(): void {
+    if (!this.playing) return;
+    this.playing = false;
+    for (const listener of this.endedListeners) listener();
+  }
+
+  private handleSocketData(chunk: Buffer | string): void {
+    this.readBuffer += chunk.toString();
+    let newline = this.readBuffer.indexOf('\n');
+    while (newline !== -1) {
+      const line = this.readBuffer.slice(0, newline).trim();
+      this.readBuffer = this.readBuffer.slice(newline + 1);
+      newline = this.readBuffer.indexOf('\n');
+      if (!line) continue;
+      let message: unknown;
+      try {
+        message = JSON.parse(line);
+      } catch {
+        continue;
+      }
+      if (
+        typeof message === 'object' &&
+        message !== null &&
+        'event' in message &&
+        message.event === 'property-change' &&
+        'name' in message &&
+        'data' in message
+      ) {
+        this.handlePropertyChange(message.name, message.data);
+      }
+    }
+  }
+
+  private handlePropertyChange(name: unknown, data: unknown): void {
+    if (name === 'eof-reached') this.eofReached = data === true;
+    else if (name === 'pause') this.paused = data === true;
+    else return;
+    if (this.playing && this.eofReached && this.paused) this.finishPlayback();
+  }
+
+  dispose(): void {
+    if (this.disposed) return;
+    this.disposed = true;
+    this.releaseResources();
+  }
+
+  private releaseResources(): void {
+    this.cancelRetryWait();
+    try {
+      this.send(['quit']);
+    } catch {
+      // The process may already have exited.
+    }
+    this.socket?.end();
+    this.socket?.destroy();
+    this.socket = null;
+    const child = this.process;
+    this.process = null;
+    child?.kill();
+    if (this.socketPath && this.deps.platform !== 'win32') {
+      try {
+        this.deps.removeSocketFile(this.socketPath);
+      } catch {
+        // mpv may still be releasing the socket. The OS temp directory owns cleanup.
+      }
+    }
+    this.socketPath = null;
+  }
+
+  private send(command: Array<string | number | boolean>): void {
+    if (!this.socket || this.socket.destroyed) {
+      throw new Error('Preview player is not connected');
+    }
+    this.socket.write(`${JSON.stringify({ command })}\n`);
+  }
+
+  private async connectWithRetry(socketPath: string): Promise<void> {
+    const deadline = this.deps.now() + CONNECT_TIMEOUT_MS;
+    while (!this.disposed && this.deps.now() < deadline) {
+      if (this.startupError) {
+        throw this.startupError;
+      }
+      try {
+        const remainingMs = deadline - this.deps.now();
+        if (remainingMs <= 0) break;
+        const socket = await this.connectOnce(
+          socketPath,
+          Math.min(CONNECT_ATTEMPT_TIMEOUT_MS, remainingMs),
+        );
+        if (this.disposed) {
+          socket.destroy();
+          throw new Error('Preview session is closed');
+        }
+        this.socket = socket;
+        this.readBuffer = '';
+        socket.on('data', (chunk: Buffer | string) => {
+          if (this.socket === socket) this.handleSocketData(chunk);
+        });
+        socket.once('close', () => this.finishPlayback());
+        this.send(['observe_property', EOF_OBSERVER_ID, 'eof-reached']);
+        this.send(['observe_property', PAUSE_OBSERVER_ID, 'pause']);
+        return;
+      } catch {
+        if (this.disposed) {
+          throw new Error('Preview session is closed');
+        }
+        const remainingMs = deadline - this.deps.now();
+        if (remainingMs <= 0) break;
+        await this.waitForRetry(Math.min(CONNECT_RETRY_MS, remainingMs));
+      }
+    }
+    if (this.startupError) {
+      throw this.startupError;
+    }
+    if (this.disposed) {
+      throw new Error('Preview session is closed');
+    }
+    throw new Error('Timed out starting the hidden mpv preview player');
+  }
+
+  private waitForRetry(delayMs: number): Promise<void> {
+    return new Promise<void>((resolve) => {
+      const timeout = this.deps.schedule(() => {
+        if (this.retryWait?.timeout === timeout) this.retryWait = null;
+        resolve();
+      }, delayMs);
+      this.retryWait = { timeout, resolve };
+    });
+  }
+
+  private cancelRetryWait(): void {
+    const pending = this.retryWait;
+    this.retryWait = null;
+    if (!pending) return;
+    this.deps.cancelSchedule(pending.timeout);
+    pending.resolve();
+  }
+
+  private connectOnce(socketPath: string, timeoutMs: number): Promise<Socket> {
+    return new Promise<Socket>((resolve, reject) => {
+      let timeout: ReturnType<typeof setTimeout> | null = null;
+      let settled = false;
+      const clearAttemptTimeout = (): void => {
+        if (timeout !== null) this.deps.cancelSchedule(timeout);
+        timeout = null;
+      };
+      const socket = this.deps.connectSocket(socketPath);
+      const onConnect = (): void => {
+        if (settled) return;
+        settled = true;
+        clearAttemptTimeout();
+        socket.off('error', onError);
+        socket.on('error', () => {
+          socket.destroy();
+          if (this.socket === socket) this.socket = null;
+        });
+        socket.once('close', () => {
+          if (this.socket === socket) this.socket = null;
+        });
+        resolve(socket);
+      };
+      const onError = (error: Error): void => {
+        if (settled) return;
+        settled = true;
+        clearAttemptTimeout();
+        socket.off('connect', onConnect);
+        socket.on('error', () => {});
+        socket.destroy();
+        reject(error);
+      };
+      socket.once('connect', onConnect);
+      socket.once('error', onError);
+      timeout = this.deps.schedule(() => {
+        if (settled) return;
+        settled = true;
+        timeout = null;
+        socket.off('connect', onConnect);
+        socket.off('error', onError);
+        socket.on('error', () => {});
+        socket.destroy();
+        reject(new Error('Timed out connecting to the hidden mpv preview player'));
+      }, timeoutMs);
+    });
+  }
+}
@@ -0,0 +1,125 @@
+import assert from 'node:assert/strict';
+import test from 'node:test';
+import {
+  buildSpeechWaveformArgs,
+  computeWaveformPeaks,
+  generateSpeechWaveform,
+} from './media-timing-waveform';
+
+function pcm(samples: number[]): Buffer {
+  const result = Buffer.alloc(samples.length * 2);
+  samples.forEach((sample, index) => result.writeInt16LE(sample, index * 2));
+  return result;
+}
+
+test('speech waveform maps the selected FFmpeg stream and visible range', () => {
+  const args = buildSpeechWaveformArgs(
+    {
+      mediaPath: '/video/show.mkv',
+      startTime: 8,
+      endTime: 15,
+      audioStreamIndex: 3,
+    },
+    'center',
+  );
+
+  assert.deepEqual(args.slice(args.indexOf('-ss'), args.indexOf('-t') + 2), [
+    '-ss',
+    '8',
+    '-i',
+    '/video/show.mkv',
+    '-t',
+    '7',
+  ]);
+  assert.deepEqual(args.slice(args.indexOf('-map'), args.indexOf('-map') + 2), ['-map', '0:3']);
+  assert.match(args[args.indexOf('-af') + 1] ?? '', /c0=FC/);
+});
+
+test('speech waveform seeks cached windows by source timestamps', () => {
+  const args = buildSpeechWaveformArgs(
+    {
+      mediaPath: { path: '/tmp/window.mkv', absoluteTimestamps: true, singleResolvedStream: true },
+      startTime: 8,
+      endTime: 15,
+    },
+    'downmix',
+  );
+
+  assert.deepEqual(args.slice(args.indexOf('-ss'), args.indexOf('-t') + 2), [
+    '-ss',
+    '8',
+    '-seek_timestamp',
+    '1',
+    '-i',
+    '/tmp/window.mkv',
+    '-t',
+    '7',
+  ]);
+  assert.equal(args.includes('-map'), false);
+});
+
+test('waveform levels rise with loudness and top out at the reference level', () => {
+  const peaks = computeWaveformPeaks(pcm([0, 1_000, -2_000, 4_000, -8_000, 16_000]), 3);
+
+  assert.equal(peaks.length, 3);
+  assert.equal(peaks[0], 0);
+  assert.ok((peaks[1] ?? 0) > 0);
+  assert.ok((peaks[1] ?? 0) < (peaks[2] ?? 0));
+  assert.equal(peaks[2], 1);
+});
+
+test('waveform flattens steady background noise and keeps speech bursts tall', () => {
+  // 20 slices of steady noise at a fixed level with an 18 dB louder "speech" burst in the middle.
+  const noise = 1_000;
+  const samples: number[] = [];
+  for (let slice = 0; slice < 20; slice += 1) {
+    const level = slice >= 8 && slice < 12 ? noise * 8 : noise;
+    for (let sample = 0; sample < 50; sample += 1) {
+      samples.push(sample % 2 === 0 ? level : -level);
+    }
+  }
+
+  const peaks = computeWaveformPeaks(pcm(samples), 20);
+
+  for (const [index, peak] of peaks.entries()) {
+    if (index >= 8 && index < 12) assert.equal(peak, 1);
+    else assert.equal(peak, 0);
+  }
+});
+
+test('waveform stays flat when the whole range is a single steady level', () => {
+  const peaks = computeWaveformPeaks(
+    pcm(Array.from({ length: 400 }, (_, i) => (i % 2 ? 900 : -900))),
+    40,
+  );
+
+  assert.ok(peaks.every((peak) => peak === 0));
+});
+
+test('speech waveform uses a mono downmix when the source has no center activity', async () => {
+  const calls: string[][] = [];
+  const peaks = await generateSpeechWaveform(
+    { mediaPath: '/video/show.mkv', startTime: 0, endTime: 2 },
+    async (args) => {
+      calls.push(args);
+      return calls.length === 1 ? pcm([0, 0, 0, 0]) : pcm([0, 4_000, -8_000, 16_000]);
+    },
+  );
+
+  assert.equal(calls.length, 2);
+  assert.match(calls[1]?.[calls[1].indexOf('-af') + 1] ?? '', /channel_layouts=mono/);
+  assert.equal(Math.max(...peaks), 1);
+});
+
+test('speech waveform keeps an active center channel without doing a second decode', async () => {
+  let calls = 0;
+  await generateSpeechWaveform(
+    { mediaPath: '/video/show.mkv', startTime: 0, endTime: 2 },
+    async () => {
+      calls += 1;
+      return pcm([0, 4_000, -8_000, 16_000]);
+    },
+  );
+
+  assert.equal(calls, 1);
+});
@@ -0,0 +1,185 @@
+import { spawn } from 'node:child_process';
+import { normalizeMediaInput, type MediaInput } from '../../media-input';
+
+const WAVEFORM_SAMPLE_RATE = 8_000;
+const WAVEFORM_POINT_COUNT = 480;
+const WAVEFORM_TIMEOUT_MS = 15_000;
+const MAX_WAVEFORM_BYTES = 16 * 1024 * 1024;
+// Keep the band where speech intelligibility lives; bass, drums, and hum sit below it.
+const SPEECH_FILTER = 'highpass=f=250,lowpass=f=3500';
+const NOISE_FLOOR_PERCENTILE = 0.2;
+const REFERENCE_PERCENTILE = 0.95;
+const NOISE_GATE_DB = 3;
+const MIN_DISPLAY_RANGE_DB = 12;
+const SILENCE_DB = -100;
+const CENTER_CHANNEL_FILTER = `pan=mono|c0=FC,${SPEECH_FILTER}`;
+const DOWNMIX_FILTER = `aformat=channel_layouts=mono,${SPEECH_FILTER}`;
+
+export interface SpeechWaveformOptions {
+  mediaPath: MediaInput;
+  startTime: number;
+  endTime: number;
+  audioStreamIndex?: number;
+}
+
+type RunFfmpeg = (args: string[]) => Promise<Buffer>;
+
+export function buildSpeechWaveformArgs(
+  options: SpeechWaveformOptions,
+  mode: 'center' | 'downmix',
+): string[] {
+  const duration = options.endTime - options.startTime;
+  const input = normalizeMediaInput(options.mediaPath);
+  const args = [
+    '-hide_banner',
+    '-nostdin',
+    '-loglevel',
+    'error',
+    '-ss',
+    String(options.startTime),
+    ...input.inputArgs,
+    '-i',
+    input.path,
+    '-t',
+    String(duration),
+  ];
+  if (
+    options.audioStreamIndex !== undefined &&
+    Number.isInteger(options.audioStreamIndex) &&
+    options.audioStreamIndex >= 0
+  ) {
+    args.push('-map', `0:${options.audioStreamIndex}`);
+  }
+  args.push(
+    '-vn',
+    '-sn',
+    '-dn',
+    '-af',
+    mode === 'center' ? CENTER_CHANNEL_FILTER : DOWNMIX_FILTER,
+    '-ac',
+    '1',
+    '-ar',
+    String(WAVEFORM_SAMPLE_RATE),
+    '-f',
+    's16le',
+    'pipe:1',
+  );
+  return args;
+}
+
+function runFfmpeg(args: string[]): Promise<Buffer> {
+  return new Promise((resolve, reject) => {
+    const child = spawn('ffmpeg', args, { stdio: ['ignore', 'pipe', 'pipe'] });
+    const chunks: Buffer[] = [];
+    let byteLength = 0;
+    let stderr = '';
+    let settled = false;
+    const timeout = setTimeout(() => {
+      if (settled) return;
+      settled = true;
+      child.kill('SIGKILL');
+      reject(new Error(`FFmpeg waveform analysis timed out after ${WAVEFORM_TIMEOUT_MS}ms`));
+    }, WAVEFORM_TIMEOUT_MS);
+
+    const settle = (callback: () => void): void => {
+      if (settled) return;
+      settled = true;
+      clearTimeout(timeout);
+      callback();
+    };
+
+    child.stdout.on('data', (chunk: Buffer) => {
+      if (settled) return;
+      byteLength += chunk.byteLength;
+      if (byteLength > MAX_WAVEFORM_BYTES) {
+        settle(() => {
+          child.kill('SIGKILL');
+          reject(new Error('The visible waveform range is too large to analyze.'));
+        });
+        return;
+      }
+      chunks.push(chunk);
+    });
+    child.stderr.setEncoding('utf8');
+    child.stderr.on('data', (chunk) => {
+      if (stderr.length < 4_000) stderr += String(chunk);
+    });
+    child.once('error', (error) => settle(() => reject(error)));
+    child.once('close', (code) => {
+      settle(() => {
+        if (code === 0) {
+          resolve(Buffer.concat(chunks, byteLength));
+          return;
+        }
+        reject(new Error(stderr.trim() || `FFmpeg exited with status ${code ?? 'unknown'}`));
+      });
+    });
+  });
+}
+
+function percentile(sortedValues: number[], fraction: number): number {
+  const index = Math.min(sortedValues.length - 1, Math.floor(sortedValues.length * fraction));
+  return sortedValues[index] ?? SILENCE_DB;
+}
+
+/**
+ * Turns mono PCM into 0..1 display heights. Each point is the RMS level of its slice in
+ * dB, measured against the clip's own noise floor (a low percentile of the slices), so
+ * constant background noise draws flat and sustained speech stands out. Peak sampling
+ * would instead follow music transients and lift the floor to nearly speech height.
+ */
+export function computeWaveformPeaks(pcm: Buffer, pointCount = WAVEFORM_POINT_COUNT): number[] {
+  const sampleCount = Math.floor(pcm.byteLength / 2);
+  if (sampleCount === 0 || pointCount <= 0) return [];
+  const resolvedPointCount = Math.min(pointCount, sampleCount);
+  const levelsDb = Array.from({ length: resolvedPointCount }, () => SILENCE_DB);
+
+  for (let point = 0; point < resolvedPointCount; point += 1) {
+    const sampleStart = Math.floor((point * sampleCount) / resolvedPointCount);
+    const sampleEnd = Math.max(
+      sampleStart + 1,
+      Math.floor(((point + 1) * sampleCount) / resolvedPointCount),
+    );
+    let energy = 0;
+    for (let sample = sampleStart; sample < sampleEnd; sample += 1) {
+      const value = pcm.readInt16LE(sample * 2) / 32_768;
+      energy += value * value;
+    }
+    const rms = Math.sqrt(energy / (sampleEnd - sampleStart));
+    levelsDb[point] = rms > 0 ? Math.max(SILENCE_DB, 20 * Math.log10(rms)) : SILENCE_DB;
+  }
+
+  const sortedLevels = [...levelsDb].sort((left, right) => left - right);
+  const floorDb = percentile(sortedLevels, NOISE_FLOOR_PERCENTILE) + NOISE_GATE_DB;
+  const referenceDb = Math.max(
+    percentile(sortedLevels, REFERENCE_PERCENTILE),
+    floorDb + MIN_DISPLAY_RANGE_DB,
+  );
+  return levelsDb.map(
+    (levelDb) =>
+      Math.round(Math.min(1, Math.max(0, (levelDb - floorDb) / (referenceDb - floorDb))) * 1_000) /
+      1_000,
+  );
+}
+
+function hasAudibleSamples(pcm: Buffer): boolean {
+  for (let offset = 0; offset + 1 < pcm.byteLength; offset += 2) {
+    if (Math.abs(pcm.readInt16LE(offset)) >= 164) return true;
+  }
+  return false;
+}
+
+export async function generateSpeechWaveform(
+  options: SpeechWaveformOptions,
+  execute: RunFfmpeg = runFfmpeg,
+): Promise<number[]> {
+  try {
+    const centerPcm = await execute(buildSpeechWaveformArgs(options, 'center'));
+    if (hasAudibleSamples(centerPcm)) return computeWaveformPeaks(centerPcm);
+  } catch {
+    // Sources without a named center channel can reject the center-only filter.
+  }
+
+  const downmixPcm = await execute(buildSpeechWaveformArgs(options, 'downmix'));
+  return computeWaveformPeaks(downmixPcm);
+}
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
sudacode	c14c690875	chore(release): prepare v0.19.6	2026-09-04 01:59:39 -07:00
sudacode	84f718043a	feat(anki): add media timing review before card creation (#203 )	2026-09-04 01:40:56 -07:00
sudacode	99266294b8	fix(tsukihime): filter releases by subtitle language (#235 )	2026-09-02 20:29:59 -07:00
sudacode	c055359be1	chore(deps): update fast-uri to 3.1.6	2026-09-02 20:16:08 -07:00
sudacode	ec5a147095	fix(subtitles): merge wrapped positioned caption rows (#234 )	2026-09-01 23:11:38 -07:00
sudacode	a0635f4360	fix(subtitles): drop ASS furigana from recorded cues (#233 )	2026-09-01 22:49:20 -07:00
sudacode	c0a78ef008	feat(anki): support Senren scene-switching field grouping (#230 )	2026-09-01 22:18:30 -07:00
sudacode	87b01155df	fix(mining): copy multi-line subtitles backward from current line (#231 )	2026-09-01 22:17:41 -07:00
sudacode	fc5c49e365	fix(subtitles): keep native secondary subtitles hidden (#232 ) - Reapply hidden visibility after secondary track changes - Relinquish suppression only after successful restoration	2026-09-01 00:49:37 -07:00
sudacode	a20269e9f5	fix(jellyfin): stop inferring subtitle delays (#227 )	2026-08-31 23:31:31 -07:00
sudacode	2ad491e95c	chore(release): prepare v0.19.5	2026-08-30 01:02:44 -07:00
sudacode	6fffcc731f	fix(youtube): keep auto captions on screen for their full span (#219 )	2026-08-29 13:18:55 -07:00
sudacode	5cc21113fd	fix(subtitles): drop ASS furigana from published cues (#224 )	2026-08-29 13:15:30 -07:00
sudacode	051140f910	fix(anki): respect configured fields for word-card enrichment (#223 )	2026-08-29 13:09:31 -07:00
sudacode	6e945f0872	fix(subtitles): keep overlapping lines that join an already active cue (#221 )	2026-08-27 23:13:29 -07:00
sudacode	c2c25c0da6	fix(anki): keep overlay progress visible through card updates (#218 )	2026-08-25 20:27:58 -07:00
sudacode	556de61756	chore(changelog): split release details into nested bullets - Update changelog generation guidance and tests - Reformat current release notes and document the new style	2026-08-25 12:39:45 -07:00
sudacode	a98fb0fddf	chore(release): prepare v0.19.4	2026-08-25 02:43:48 -07:00
sudacode	e816b3b371	fix(subtitles): recover positioned ASS word spacing and drop control debris (#217 )	2026-08-25 02:22:21 -07:00
sudacodeandsudacode	ed7d3f4c3d	feat(release): track prerelease deltas and validate committed notes (#216 ) Co-authored-by: sudacode <claude@lmaoxd.lol>	2026-08-23 15:24:18 -07:00
sudacode	509dc5bf7f	fix(subtitles): improve secondary subtitle extraction and display (#215 )	2026-08-23 01:56:49 -07:00
sudacode	0a0aa3ec98	chore(release): prepare v0.19.4-beta.4	2026-08-23 01:23:16 -07:00
sudacode	b87cc3cfdd	chore(workflow): add release skill	2026-08-23 01:14:03 -07:00
sudacode	8cb3c8c90a	fix(macos): pin window helper deployment target to fix older systems - Build the Swift window-tracking helper with an explicit `-target ...-apple-macos12.0` instead of letting swiftc stamp the build machine's OS, so the overlay attaches to mpv on older macOS (e.g. Ventura) instead of crashing at load - Fold the standalone build-macos-helper.sh into prepare-build-assets.mjs and update docs/tests to match	2026-08-23 00:58:24 -07:00
sudacode	03ea903927	fix(subtitles): preserve authored spaces in fragmented ASS karaoke - Only trim indentation before slicing ASS event fields so a trailing authored space at a fragment's end is kept, preventing words from being joined together when fragmented karaoke lines are reconstructed - Add regression test covering event-boundary word spacing	2026-08-22 23:32:35 -07:00
sudacode	3aea42e6f8	chore(release): prepare v0.19.4-beta.3	2026-08-21 23:25:37 -07:00
sudacode	88bb3edfa4	fix(subtitles): preserve lyric selection and deduplicate secondary text - Seek sidebar selections past overlapping karaoke exit spans - Collapse duplicate long ASS lines and omit reconstructed sign grids	2026-08-21 01:28:00 -07:00
sudacode	9445aef004	fix(subtitles): navigate by sanitized cues across ASS animation - Route subtitle navigation through parsed cue boundaries - Treat explicit short seeks as subtitle priming events	2026-08-20 00:20:51 -07:00
sudacode	c01bcd9d0f	fix(subtitles): collapse layered ASS lyrics across startup and seeks - Reconstruct fragment-only karaoke while preserving canonical animated signs - Reprocess live subtitles after initial playback and seek updates - Add a development launcher for playable media files	2026-08-19 22:55:10 -07:00