chore(release): prepare v0.19.6

- Reapply hidden visibility after secondary track changes
- Relinquish suppression only after successful restoration

CHANGELOG.md

@@ -1,5 +1,36 @@
 # 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

bun.lock

@@ -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=="],
 

changes/anki-update-notifications.md

@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
-- Dismissed active overlay card-update progress when notification settings switch to OSD before an update finishes.

changes/media-timing-review.md

@@ -1,5 +0,0 @@
-type: added
-area: mining
-
-- Added optional pre-generation timing review for word, sentence, and audio cards with a compact speech-weighted waveform, clearly labeled mined-line boundaries, drag and keyboard adjustments, audio preview with a sweeping playhead, exact screenshot and AVIF timing, cancellation choices that include keeping a card without media, and a session-only runtime toggle.
-- The timing review can pull any number of previous and next subtitle lines onto the card: `P`/`N` (or the Prev/Next steppers) add lines one at a time, Shift removes them, the sentence preview highlights exactly what the card will contain, and the clip range and line boundary markers on the waveform follow the added lines automatically.

config.example.jsonc

@@ -523,7 +523,7 @@
   // ==========================================
   // AnkiConnect Integration
   // Automatic Anki updates and media generation options.
-  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume/reviewTiming, 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.
   // ==========================================
@@ -607,6 +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.

docs-site/anki-integration.md

@@ -158,6 +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.
@@ -181,7 +183,7 @@ 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. 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, 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 speech-band mono mix. 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.
+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.
 
@@ -315,9 +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": {
@@ -329,6 +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.
@@ -339,14 +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                             |

docs-site/changelog.md

@@ -1,5 +1,36 @@
 # 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**

docs-site/configuration.md

@@ -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**
 
@@ -1054,6 +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.
@@ -1083,6 +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

docs-site/jellyfin-integration.md

@@ -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
 

docs-site/mining-workflow.md

@@ -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
 

docs-site/public/config.example.jsonc

@@ -523,7 +523,7 @@
   // ==========================================
   // AnkiConnect Integration
   // Automatic Anki updates and media generation options.
-  // Hot-reload: ankiConnect.ai.enabled, media.normalizeAudio/mirrorMpvVolume/reviewTiming, 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.
   // ==========================================
@@ -607,6 +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.

docs-site/shortcuts.md

@@ -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
 

docs-site/tsukihime-integration.md

@@ -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.

docs/architecture/subtitle-overlay-priming.md

@@ -131,7 +131,19 @@ coming and prefetching would otherwise idle for the rest of the cue.
   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.
+  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

package.json

@@ -2,7 +2,7 @@
   "name": "subminer",
   "productName": "SubMiner",
   "desktopName": "SubMiner.desktop",
-  "version": "0.19.5",
+  "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",
@@ -87,7 +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",

release/release-notes.md

@@ -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`.

src/anki-integration.test.ts

@@ -155,6 +155,7 @@ function createFieldGroupingMergeCollaborator(options?: {
     getEffectiveSentenceCardConfig: () => ({
       sentenceField: 'Sentence',
       audioField: 'SentenceAudio',
+      fieldGroupingProvider: 'kiku' as const,
     }),
     getCurrentSubtitleText: () => options?.currentSubtitleText,
     resolveFieldName,

src/anki-integration.ts

@@ -852,6 +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;
@@ -860,10 +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,
@@ -871,8 +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),
     };
   }
@@ -891,7 +928,7 @@ export class AnkiIntegration {
 
   private async processNewCard(
     noteId: number,
-    options?: { skipKikuFieldGrouping?: boolean },
+    options?: { skipFieldGrouping?: boolean },
   ): Promise<void> {
     await this.noteUpdateWorkflow.execute(noteId, options);
   }
@@ -1521,7 +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;
     }
 

src/anki-integration/card-creation-manual-update.test.ts

@@ -125,8 +125,7 @@ function createManualUpdateService(overrides: Partial<CardCreationDeps> = {}): {
       audioField: 'SentenceAudio',
       lapisEnabled: false,
       kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
     getFallbackDurationSeconds: () => 10,
     appendKnownWordsFromNoteInfo: () => undefined,
@@ -211,8 +210,7 @@ test('manual clipboard word-card update uses configured fields with Lapis and Ki
       audioField: 'SentenceAudio',
       lapisEnabled: true,
       kikuEnabled: true,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
   });
 
@@ -279,8 +277,7 @@ test('audio-card action keeps Lapis and Kiku sentence fields', async () => {
       audioField: 'SentenceAudio',
       lapisEnabled: true,
       kikuEnabled: true,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
   });
 
@@ -343,8 +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,
   });

src/anki-integration/card-creation-sentence-media.test.ts

@@ -120,8 +120,7 @@ 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,

src/anki-integration/card-creation.test.ts

@@ -70,8 +70,7 @@ test('CardCreationService counts locally created sentence cards', async () => {
       audioField: 'SentenceAudio',
       lapisEnabled: false,
       kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
     getFallbackDurationSeconds: () => 10,
     appendKnownWordsFromNoteInfo: () => undefined,
@@ -171,8 +170,7 @@ test('CardCreationService keeps updating after trackLastAddedNoteId throws', asy
       audioField: 'SentenceAudio',
       lapisEnabled: false,
       kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
     getFallbackDurationSeconds: () => 10,
     appendKnownWordsFromNoteInfo: () => undefined,
@@ -272,8 +270,7 @@ test('CardCreationService keeps updating after recordCardsMinedCallback throws',
       audioField: 'SentenceAudio',
       lapisEnabled: false,
       kikuEnabled: false,
-      kikuFieldGrouping: 'disabled',
-      kikuDeleteDuplicateInAuto: false,
+      fieldGroupingMode: 'disabled',
     }),
     getFallbackDurationSeconds: () => 10,
     appendKnownWordsFromNoteInfo: () => undefined,
@@ -394,8 +391,7 @@ 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,
@@ -499,8 +495,7 @@ 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,
@@ -640,8 +635,7 @@ 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,
@@ -741,8 +735,7 @@ 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,
@@ -832,8 +825,7 @@ 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,

src/anki-integration/card-creation.ts

@@ -139,8 +139,7 @@ interface CardCreationDeps {
     audioField: string;
     lapisEnabled: boolean;
     kikuEnabled: boolean;
-    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
-    kikuDeleteDuplicateInAuto: boolean;
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
     wordCardKind?: WordCardKind;
   };
   getFallbackDurationSeconds: () => number;
@@ -697,8 +696,7 @@ export class CardCreationService {
         ).trim();
         let duplicateNoteIds: number[] = [];
         if (
-          sentenceCardConfig.kikuEnabled &&
-          sentenceCardConfig.kikuFieldGrouping !== 'disabled' &&
+          sentenceCardConfig.fieldGroupingMode !== 'disabled' &&
           pendingExpressionText &&
           this.deps.findDuplicateNoteIds
         ) {

src/anki-integration/field-grouping-merge.test.ts

@@ -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>');
+});

src/anki-integration/field-grouping-merge.ts

@@ -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);

src/anki-integration/field-grouping-workflow.test.ts

@@ -71,7 +71,7 @@ function createWorkflowHarness() {
     getEffectiveSentenceCardConfig: () => ({
       sentenceField: 'Sentence',
       audioField: 'SentenceAudio',
-      kikuDeleteDuplicateInAuto: true,
+      fieldGroupingDeleteDuplicateInAuto: true,
     }),
     getCurrentSubtitleText: () => 'subtitle-text',
     getFieldGroupingCallback: (): FieldGroupingCallback | null => {

src/anki-integration/field-grouping-workflow.ts

@@ -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);

src/anki-integration/field-grouping.test.ts

@@ -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(),

src/anki-integration/field-grouping.ts

@@ -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,

src/anki-integration/note-update-workflow.test.ts

@@ -59,7 +59,7 @@ function createWorkflowHarness() {
       sentenceField: 'Sentence',
       lapisEnabled: false,
       kikuEnabled: false,
-      kikuFieldGrouping: 'disabled' as const,
+      fieldGroupingMode: 'disabled' as const,
     }),
     appendKnownWordsFromNoteInfo: (_noteInfo: NoteUpdateWorkflowNoteInfo) => undefined,
     removeKnownWordNote: (_noteId: number) => undefined,
@@ -138,7 +138,7 @@ test('NoteUpdateWorkflow uses configured fields for word-card enrichment with La
     sentenceField: 'Sentence',
     lapisEnabled: true,
     kikuEnabled: true,
-    kikuFieldGrouping: 'disabled',
+    fieldGroupingMode: 'disabled',
   });
   harness.deps.client.notesInfo = async () =>
     [
@@ -195,7 +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 () =>
     [
@@ -228,7 +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 () =>
@@ -264,7 +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 () =>
@@ -319,7 +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 () =>
     [
@@ -362,7 +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 () => {

src/anki-integration/note-update-workflow.ts

@@ -46,7 +46,7 @@ export interface NoteUpdateWorkflowDeps {
     sentenceField: string;
     lapisEnabled: boolean;
     kikuEnabled: boolean;
-    kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
+    fieldGroupingMode: 'auto' | 'manual' | 'disabled';
     wordCardKind?: WordCardKind;
   };
   appendKnownWordsFromNoteInfo: (noteInfo: NoteUpdateWorkflowNoteInfo) => void;
@@ -170,7 +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]);
@@ -196,9 +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);
@@ -401,7 +399,7 @@ export class NoteUpdateWorkflow {
           noteInfoForGrouping = refreshedInfo[0]!;
         }
 
-        if (sentenceCardConfig.kikuFieldGrouping === 'auto') {
+        if (sentenceCardConfig.fieldGroupingMode === 'auto') {
           await this.deps.handleFieldGroupingAuto(
             duplicateNoteId,
             noteId,
@@ -410,7 +408,7 @@ export class NoteUpdateWorkflow {
           );
           return;
         }
-        if (sentenceCardConfig.kikuFieldGrouping === 'manual') {
+        if (sentenceCardConfig.fieldGroupingMode === 'manual') {
           await this.deps.handleFieldGroupingManual(
             duplicateNoteId,
             noteId,

src/anki-integration/runtime.ts

@@ -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 }

src/config/config.test.ts

@@ -2189,6 +2189,7 @@ test('runtime options registry is centralized', () => {
     'subtitle.annotation.frequency',
     'anki.nPlusOneMatchMode',
     'anki.kikuFieldGrouping',
+    'anki.senrenFieldGrouping',
   ]);
 });
 
@@ -2776,6 +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(

src/config/definitions/defaults-integrations.ts

@@ -92,6 +92,11 @@ export const INTEGRATIONS_DEFAULT_CONFIG: Pick<
       fieldGrouping: 'disabled',
       deleteDuplicateInAuto: true,
     },
+    isSenren: {
+      enabled: false,
+      fieldGrouping: 'auto',
+      deleteDuplicateInAuto: true,
+    },
     lapisKiku: {
       wordCardKind: 'word-and-sentence',
     },

src/config/definitions/options-integrations.ts

@@ -371,6 +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',

src/config/definitions/runtime-options.ts

@@ -138,5 +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',
+        },
+      }),
+    },
   ];
 }

src/config/definitions/template-sections.ts

@@ -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/reviewTiming, 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.',
     ],

src/config/resolve/anki-connect.ts

@@ -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);
 }

src/config/resolve/anki-connect/initialize.ts

@@ -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,
     },

src/config/resolve/anki-connect/senren.ts

@@ -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;
+  }
+}

src/config/settings/registry.test.ts

@@ -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'],
   );
 });
 
@@ -367,6 +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);
   }

src/config/settings/registry.ts

@@ -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',
@@ -252,7 +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':
@@ -408,9 +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' };
@@ -711,6 +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' ||

src/core/services/config-hot-reload.ts

@@ -86,6 +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;
 

src/core/services/immersion-tracker/lexical-rollups.test.ts

@@ -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);
 

src/core/services/index.ts

@@ -131,12 +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,

src/core/services/jellyfin-subtitle-delay.test.ts

@@ -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,
-  );
-});

src/core/services/jellyfin-subtitle-delay.ts

@@ -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;
-  }
-}

src/core/services/media-timing-preview.test.ts

@@ -23,6 +23,21 @@ describe('buildMediaTimingPreviewArgs', () => {
     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',
@@ -189,3 +204,88 @@ test('preview session bounds a connection attempt that never settles', async ()
   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();
+});

src/core/services/media-timing-preview.ts

@@ -8,12 +8,21 @@ 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'>;
@@ -51,6 +60,9 @@ export function buildMediaTimingPreviewArgs(
   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;
 }
@@ -89,6 +101,11 @@ export class MediaTimingPreviewSession {
     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 = {
@@ -155,6 +172,11 @@ export class MediaTimingPreviewSession {
     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');
@@ -163,18 +185,68 @@ export class MediaTimingPreviewSession {
       throw new Error('Preview timing is invalid');
     }
 
+    this.playing = false;
     this.send(['set_property', 'pause', true]);
-    this.send(['set_property', 'ab-loop-a', startTime]);
-    this.send(['set_property', 'ab-loop-b', endTime]);
     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;
@@ -229,6 +301,13 @@ export class MediaTimingPreviewSession {
           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) {

src/core/services/media-timing-waveform.test.ts

@@ -35,16 +35,67 @@ test('speech waveform maps the selected FFmpeg stream and visible range', () =>
   assert.match(args[args.indexOf('-af') + 1] ?? '', /c0=FC/);
 });
 
-test('waveform peaks are normalized without flattening quieter sections', () => {
+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.ok((peaks[0] ?? 0) > 0);
-  assert.ok((peaks[0] ?? 0) < (peaks[1] ?? 0));
+  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(

src/core/services/media-timing-waveform.ts

@@ -1,15 +1,22 @@
 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;
-const SPEECH_FILTER = 'highpass=f=120,lowpass=f=4000';
+// 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: string;
+  mediaPath: MediaInput;
   startTime: number;
   endTime: number;
   audioStreamIndex?: number;
@@ -22,6 +29,7 @@ export function buildSpeechWaveformArgs(
   mode: 'center' | 'downmix',
 ): string[] {
   const duration = options.endTime - options.startTime;
+  const input = normalizeMediaInput(options.mediaPath);
   const args = [
     '-hide_banner',
     '-nostdin',
@@ -29,8 +37,9 @@ export function buildSpeechWaveformArgs(
     'error',
     '-ss',
     String(options.startTime),
+    ...input.inputArgs,
     '-i',
-    options.mediaPath,
+    input.path,
     '-t',
     String(duration),
   ];
@@ -108,11 +117,22 @@ function runFfmpeg(args: string[]): Promise<Buffer> {
   });
 }
 
+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 peaks = Array.from({ length: resolvedPointCount }, () => 0);
+  const levelsDb = Array.from({ length: resolvedPointCount }, () => SILENCE_DB);
 
   for (let point = 0; point < resolvedPointCount; point += 1) {
     const sampleStart = Math.floor((point * sampleCount) / resolvedPointCount);
@@ -120,18 +140,25 @@ export function computeWaveformPeaks(pcm: Buffer, pointCount = WAVEFORM_POINT_CO
       sampleStart + 1,
       Math.floor(((point + 1) * sampleCount) / resolvedPointCount),
     );
-    let peak = 0;
+    let energy = 0;
     for (let sample = sampleStart; sample < sampleEnd; sample += 1) {
-      peak = Math.max(peak, Math.abs(pcm.readInt16LE(sample * 2)) / 32_768);
+      const value = pcm.readInt16LE(sample * 2) / 32_768;
+      energy += value * value;
     }
-    peaks[point] = peak;
+    const rms = Math.sqrt(energy / (sampleEnd - sampleStart));
+    levelsDb[point] = rms > 0 ? Math.max(SILENCE_DB, 20 * Math.log10(rms)) : SILENCE_DB;
   }
 
-  const sortedPeaks = [...peaks].sort((left, right) => left - right);
-  const referenceIndex = Math.min(sortedPeaks.length - 1, Math.floor(sortedPeaks.length * 0.95));
-  const referencePeak = Math.max(sortedPeaks[referenceIndex] ?? 0, 0.01);
-  return peaks.map(
-    (peak) => Math.round(Math.sqrt(Math.min(1, peak / referencePeak)) * 1_000) / 1_000,
+  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,
   );
 }
 

src/core/services/mining.test.ts

@@ -244,6 +244,35 @@ test('handleMultiCopyDigit copies available history and reports truncation', ()
   assert.equal(osd.at(-1), 'Only 2 lines available, copied 2');
 });
 
+test('handleMultiCopyDigit copies backward from the current subtitle after a backward seek', () => {
+  const copied: string[] = [];
+  const tracker = new SubtitleTimingTracker();
+
+  try {
+    tracker.recordSubtitle('A', 1, 2);
+    tracker.recordSubtitle('B', 3, 4);
+    tracker.recordSubtitle('C', 5, 6);
+    tracker.recordSubtitle('B', 3, 4);
+
+    const deps = {
+      subtitleTimingTracker: tracker,
+      writeClipboardText: (text: string) => copied.push(text),
+      showMpvOsd: () => {},
+    };
+
+    handleMultiCopyDigit(1, deps);
+    handleMultiCopyDigit(2, deps);
+
+    assert.deepEqual(copied, ['B', 'A\n\nB']);
+    assert.deepEqual(tracker.getRecentEntries(2), [
+      { displayText: 'A', startTime: 1, endTime: 2, secondaryText: undefined },
+      { displayText: 'B', startTime: 3, endTime: 4, secondaryText: undefined },
+    ]);
+  } finally {
+    tracker.destroy();
+  }
+});
+
 test('handleMineSentenceDigit reports async create failures', async () => {
   const osd: string[] = [];
   const logs: Array<{ message: string; err: unknown }> = [];
@@ -344,6 +373,22 @@ test('handleMineSentenceDigit keeps per-entry timings when subtitle text repeats
   }
 });
 
+test('subtitle timing history preserves adjacent repeated text with distinct timings', () => {
+  const tracker = new SubtitleTimingTracker();
+
+  try {
+    tracker.recordSubtitle('same', 1, 2);
+    tracker.recordSubtitle('same', 3, 4);
+
+    assert.deepEqual(tracker.getRecentEntries(2), [
+      { displayText: 'same', startTime: 1, endTime: 2, secondaryText: undefined },
+      { displayText: 'same', startTime: 3, endTime: 4, secondaryText: undefined },
+    ]);
+  } finally {
+    tracker.destroy();
+  }
+});
+
 test('handleMineSentenceDigit joins per-entry secondary subtitles when available', async () => {
   const created: Array<{ sentence: string; secondarySub?: string }> = [];
   const tracker = new SubtitleTimingTracker();

src/core/services/mpv-protocol.test.ts

@@ -83,6 +83,7 @@ function createDeps(overrides: Partial<MpvProtocolHandleMessageDeps> = {}): {
         state.secondarySubText = text;
       },
       resolvePendingRequest: () => false,
+      shouldEnforceSecondarySubVisibilityHidden: () => true,
       setSecondarySubVisibility: () => {},
       syncCurrentAudioStreamIndex: () => {},
       setCurrentAudioTrackId: () => {},
@@ -198,6 +199,21 @@ test('dispatchMpvProtocolMessage rejects decimal subtitle track IDs', async () =
   assert.deepEqual(state.events, [{ sid: null }, { sid: null }, { sid: null }, { sid: null }]);
 });
 
+test('dispatchMpvProtocolMessage hides native secondary subtitles after a track change', async () => {
+  const visibilityChanges: boolean[] = [];
+  const { deps, state } = createDeps({
+    setSecondarySubVisibility: (visible) => visibilityChanges.push(visible),
+  });
+
+  await dispatchMpvProtocolMessage(
+    { event: 'property-change', name: 'secondary-sid', data: '4' },
+    deps,
+  );
+
+  assert.deepEqual(visibilityChanges, [false]);
+  assert.deepEqual(state.events, [{ sid: 4 }]);
+});
+
 test('dispatchMpvProtocolMessage enforces sub-visibility hidden when overlay suppression is enabled', async () => {
   const { deps, state } = createDeps({
     isVisibleOverlayVisible: () => true,
@@ -239,6 +255,24 @@ test('dispatchMpvProtocolMessage skips sub-visibility suppression when overlay i
   assert.equal(state.commands.length, 0);
 });
 
+test('dispatchMpvProtocolMessage corrects native secondary subtitle visibility', async () => {
+  const visibilityChanges: boolean[] = [];
+  const { deps } = createDeps({
+    setSecondarySubVisibility: (visible) => visibilityChanges.push(visible),
+  });
+
+  await dispatchMpvProtocolMessage(
+    { event: 'property-change', name: 'secondary-sub-visibility', data: 'yes' },
+    deps,
+  );
+  await dispatchMpvProtocolMessage(
+    { event: 'property-change', name: 'secondary-sub-visibility', data: 'no' },
+    deps,
+  );
+
+  assert.deepEqual(visibilityChanges, [false]);
+});
+
 test('dispatchMpvProtocolMessage sets secondary subtitle track based on track list response', async () => {
   const { deps, state } = createDeps();
 

src/core/services/mpv-protocol.ts

@@ -72,6 +72,7 @@ export interface MpvProtocolHandleMessageDeps {
   emitSubtitleMetricsChange: (payload: Partial<MpvSubtitleRenderMetrics>) => void;
   setCurrentSecondarySubText: (text: string) => void;
   resolvePendingRequest: (requestId: number, message: MpvMessage) => boolean;
+  shouldEnforceSecondarySubVisibilityHidden: () => boolean;
   setSecondarySubVisibility: (visible: boolean) => void;
   syncCurrentAudioStreamIndex: () => void;
   setCurrentAudioTrackId: (value: number | null) => void;
@@ -285,6 +286,9 @@ export async function dispatchMpvProtocolMessage(
             : null;
       deps.emitSubtitleTrackChange({ sid: sid !== null && Number.isInteger(sid) ? sid : null });
     } else if (msg.name === 'secondary-sid') {
+      if (deps.shouldEnforceSecondarySubVisibilityHidden()) {
+        deps.setSecondarySubVisibility(false);
+      }
       const sid =
         typeof msg.data === 'number'
           ? msg.data
@@ -375,6 +379,11 @@ export async function dispatchMpvProtocolMessage(
       if (deps.isVisibleOverlayVisible() && asBoolean(msg.data, false)) {
         deps.sendCommand({ command: ['set_property', 'sub-visibility', false] });
       }
+    } else if (msg.name === 'secondary-sub-visibility') {
+      const visible = parseVisibilityProperty(msg.data);
+      if (deps.shouldEnforceSecondarySubVisibilityHidden() && visible === true) {
+        deps.setSecondarySubVisibility(false);
+      }
     } else if (msg.name === 'sub-use-margins') {
       deps.emitSubtitleMetricsChange({
         subUseMargins: asBoolean(msg.data, deps.getSubtitleMetrics().subUseMargins),

src/core/services/mpv.test.ts

@@ -652,7 +652,7 @@ test('MpvIpcClient captures and disables secondary subtitle visibility on reques
   ]);
 });
 
-test('MpvIpcClient restorePreviousSecondarySubVisibility restores and clears tracked value', async () => {
+test('MpvIpcClient restores secondary subtitle visibility and relinquishes suppression', async () => {
   const commands: unknown[] = [];
   const client = new MpvIpcClient('/tmp/mpv.sock', makeDeps());
   const previous: boolean[] = [];
@@ -671,6 +671,12 @@ test('MpvIpcClient restorePreviousSecondarySubVisibility restores and clears tra
   });
   client.restorePreviousSecondarySubVisibility();
 
+  await invokeHandleMessage(client, {
+    event: 'property-change',
+    name: 'secondary-sub-visibility',
+    data: 'yes',
+  });
+
   assert.equal(previous[0], true);
   assert.equal(previous.length, 1);
   assert.deepEqual(commands, [
@@ -682,8 +688,53 @@ test('MpvIpcClient restorePreviousSecondarySubVisibility restores and clears tra
     },
   ]);
 
+  await invokeHandleMessage(client, {
+    event: 'property-change',
+    name: 'secondary-sub-visibility',
+    data: 'yes',
+  });
+  assert.equal(commands.length, 2);
+
   client.restorePreviousSecondarySubVisibility();
   assert.equal(commands.length, 2);
+
+  const callbacks = (client as any).transport.callbacks;
+  callbacks.onConnect();
+  commands.length = 0;
+
+  await invokeHandleMessage(client, {
+    event: 'property-change',
+    name: 'secondary-sub-visibility',
+    data: 'yes',
+  });
+  assert.deepEqual(commands, [{ command: ['set_property', 'secondary-sub-visibility', 'no'] }]);
+});
+
+test('MpvIpcClient keeps secondary subtitle suppression when restoration send fails', async () => {
+  const commands: unknown[] = [];
+  const client = new MpvIpcClient('/tmp/mpv.sock', makeDeps());
+
+  (client as any).send = (payload: unknown) => {
+    commands.push(payload);
+    return false;
+  };
+
+  await invokeHandleMessage(client, {
+    request_id: MPV_REQUEST_ID_SECONDARY_SUB_VISIBILITY,
+    data: 'yes',
+  });
+  client.restorePreviousSecondarySubVisibility();
+  await invokeHandleMessage(client, {
+    event: 'property-change',
+    name: 'secondary-sid',
+    data: 4,
+  });
+
+  assert.deepEqual(commands, [
+    { command: ['set_property', 'secondary-sub-visibility', 'no'] },
+    { command: ['set_property', 'secondary-sub-visibility', 'yes'] },
+    { command: ['set_property', 'secondary-sub-visibility', 'no'] },
+  ]);
 });
 
 test('MpvIpcClient updates current audio stream index from track list', async () => {

src/core/services/mpv.ts

@@ -184,6 +184,7 @@ export class MpvIpcClient implements MpvClient {
     osdDimensions: null,
   };
   private previousSecondarySubVisibility: boolean | null = null;
+  private enforceSecondarySubVisibilityHidden = true;
   private playbackPaused: boolean | null = null;
   private pauseAtTime: number | null = null;
   private pendingPauseAtSubEnd = false;
@@ -199,6 +200,7 @@ export class MpvIpcClient implements MpvClient {
       socketFactory: deps.socketFactory,
       connectTimeoutMs: deps.connectTimeoutMs,
       onConnect: () => {
+        this.enforceSecondarySubVisibilityHidden = true;
         this.connected = true;
         this.connecting = false;
         this.socket = this.transport.getSocket();
@@ -476,6 +478,7 @@ export class MpvIpcClient implements MpvClient {
       },
       resolvePendingRequest: (requestId: number, message: MpvMessage) =>
         this.tryResolvePendingRequest(requestId, message),
+      shouldEnforceSecondarySubVisibilityHidden: () => this.enforceSecondarySubVisibilityHidden,
       setSecondarySubVisibility: (visible: boolean) => this.setSecondarySubVisibility(visible),
       syncCurrentAudioStreamIndex: () => {
         this.syncCurrentAudioStreamIndex();
@@ -647,9 +650,11 @@ export class MpvIpcClient implements MpvClient {
   restorePreviousSecondarySubVisibility(): void {
     const previous = this.previousSecondarySubVisibility;
     if (previous === null) return;
-    this.send({
+    const restored = this.send({
       command: ['set_property', 'secondary-sub-visibility', previous ? 'yes' : 'no'],
     });
+    if (!restored) return;
+    this.enforceSecondarySubVisibilityHidden = false;
     this.previousSecondarySubVisibility = null;
   }
 

src/core/services/remote-media-window-cache.test.ts

@@ -0,0 +1,253 @@
+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 {
+  buildRemoteMediaWindowArgs,
+  RemoteMediaWindowCache,
+  REMOTE_MEDIA_WINDOW_MAX_SECONDS,
+  type RemoteMediaWindowCacheOptions,
+} from './remote-media-window-cache';
+
+const SOURCE = {
+  path: 'https://jellyfin.example/Videos/abc/stream?static=true',
+  audioStreamIndex: 2,
+};
+
+type ExecFileStub = NonNullable<RemoteMediaWindowCacheOptions['execFile']>;
+
+function createStub(options: { fail?: boolean; empty?: boolean; defer?: boolean } = {}) {
+  const calls: string[][] = [];
+  const pendingCallbacks: Array<() => void> = [];
+  const execFile: ExecFileStub = (_file, args, _options, callback) => {
+    calls.push([...args]);
+    const finish = (): void => {
+      const outputPath = args.at(-1);
+      assert.ok(outputPath);
+      if (options.fail) {
+        callback(Object.assign(new Error('boom'), { code: 1 }));
+        return;
+      }
+      if (!options.empty) {
+        fs.writeFileSync(outputPath, 'mkv', 'utf8');
+      }
+      callback(null);
+    };
+    if (options.defer) {
+      pendingCallbacks.push(finish);
+    } else {
+      queueMicrotask(finish);
+    }
+  };
+  return {
+    calls,
+    execFile,
+    flush: () => {
+      for (const finish of pendingCallbacks.splice(0)) finish();
+    },
+  };
+}
+
+async function withCache(
+  stubOptions: Parameters<typeof createStub>[0],
+  cacheOptions: Omit<RemoteMediaWindowCacheOptions, 'execFile' | 'tempDir'>,
+  run: (cache: RemoteMediaWindowCache, stub: ReturnType<typeof createStub>) => Promise<void>,
+): Promise<void> {
+  const tempDir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-media-window-test-'));
+  const stub = createStub(stubOptions);
+  const cache = new RemoteMediaWindowCache({
+    tempDir,
+    execFile: stub.execFile,
+    idleTtlMs: 0,
+    logDebug: () => undefined,
+    ...cacheOptions,
+  });
+  try {
+    await run(cache, stub);
+  } finally {
+    cache.cleanup();
+    fs.rmSync(tempDir, { recursive: true, force: true });
+  }
+}
+
+function argValue(args: string[], flag: string): string | undefined {
+  const index = args.indexOf(flag);
+  return index === -1 ? undefined : args[index + 1];
+}
+
+test('buildRemoteMediaWindowArgs stream-copies the window with source timestamps intact', () => {
+  const args = buildRemoteMediaWindowArgs(
+    { ...SOURCE, inputOptions: { reconnect: true, headers: { Referer: 'https://a.example/' } } },
+    { startTime: 22.75, endTime: 33 },
+    '/tmp/window.mkv',
+  );
+
+  const inputIndex = args.indexOf('-i');
+  assert.equal(args[inputIndex + 1], SOURCE.path);
+  assert.ok(args.indexOf('-reconnect') < inputIndex);
+  assert.ok(args.indexOf('-headers') < inputIndex);
+  assert.equal(argValue(args, '-ss'), '22.75');
+  assert.equal(argValue(args, '-t'), '10.25');
+  assert.ok(args.indexOf('-t') < inputIndex);
+  assert.deepEqual(args.slice(args.indexOf('-map'), args.indexOf('-map') + 4), [
+    '-map',
+    '0:v:0?',
+    '-map',
+    '0:2',
+  ]);
+  assert.equal(argValue(args, '-c'), 'copy');
+  assert.ok(args.includes('-copyts'));
+  assert.ok(args.includes('-start_at_zero'));
+  assert.equal(argValue(args, '-f'), 'matroska');
+  assert.equal(args.at(-1), '/tmp/window.mkv');
+});
+
+test('buildRemoteMediaWindowArgs keeps every audio stream when none is selected', () => {
+  const args = buildRemoteMediaWindowArgs(
+    { path: SOURCE.path, audioStreamIndex: null },
+    { startTime: 0, endTime: 5 },
+    '/tmp/window.mkv',
+  );
+
+  assert.equal(args[args.lastIndexOf('-map') + 1], '0:a');
+});
+
+test('acquire downloads once and reuses the window for covered ranges', async () => {
+  await withCache({}, {}, async (cache, stub) => {
+    const window = await cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+
+    assert.equal(stub.calls.length, 1);
+    assert.equal(argValue(stub.calls[0]!, '-ss'), '9.75');
+    assert.equal(argValue(stub.calls[0]!, '-t'), '5.25');
+    assert.equal(window.startTime, 9.75);
+    assert.equal(window.endTime, 15);
+    assert.equal(window.audioStreamIndex, 2);
+    assert.ok(fs.existsSync(window.path));
+    assert.deepEqual(window.media, {
+      path: window.path,
+      source: 'remote-window',
+      singleResolvedStream: true,
+      absoluteTimestamps: true,
+    });
+
+    assert.equal(await cache.acquire(SOURCE, { startTime: 11, endTime: 15 }), window);
+    assert.equal(await cache.lookup(SOURCE, { startTime: 12, endTime: 12 }), window);
+    assert.equal(
+      await cache.lookup(
+        { path: SOURCE.path, audioStreamIndex: null },
+        { startTime: 12, endTime: 13 },
+      ),
+      window,
+    );
+    assert.equal(stub.calls.length, 1);
+  });
+});
+
+test('lookup never downloads and misses on other ranges, sources, or audio streams', async () => {
+  await withCache({}, {}, async (cache, stub) => {
+    assert.equal(await cache.lookup(SOURCE, { startTime: 10, endTime: 14 }), null);
+    assert.equal(stub.calls.length, 0);
+
+    await cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+    assert.equal(await cache.lookup(SOURCE, { startTime: 14, endTime: 16 }), null);
+    assert.equal(
+      await cache.lookup(
+        { path: 'https://other.example/stream', audioStreamIndex: 2 },
+        {
+          startTime: 11,
+          endTime: 12,
+        },
+      ),
+      null,
+    );
+    assert.equal(
+      await cache.lookup(
+        { path: SOURCE.path, audioStreamIndex: 3 },
+        { startTime: 11, endTime: 12 },
+      ),
+      null,
+    );
+    assert.equal(stub.calls.length, 1);
+  });
+});
+
+test('acquire widens to the union of the old window and replaces the old file', async () => {
+  await withCache({}, {}, async (cache, stub) => {
+    const first = await cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+    const second = await cache.acquire(SOURCE, { startTime: 8, endTime: 12 });
+
+    assert.equal(stub.calls.length, 2);
+    assert.equal(argValue(stub.calls[1]!, '-ss'), '7.75');
+    assert.equal(second.startTime, 7.75);
+    assert.equal(second.endTime, 15);
+    assert.notEqual(second.path, first.path);
+    assert.equal(fs.existsSync(first.path), false);
+    assert.ok(fs.existsSync(second.path));
+    assert.equal(cache.currentWindow, second);
+  });
+});
+
+test('acquire shares an in-flight download between concurrent callers', async () => {
+  await withCache({ defer: true }, {}, async (cache, stub) => {
+    const first = cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+    await Promise.resolve();
+    const second = cache.acquire(SOURCE, { startTime: 11, endTime: 13 });
+    const lookup = cache.lookup(SOURCE, { startTime: 12, endTime: 12 });
+    await Promise.resolve();
+    assert.equal(stub.calls.length, 1);
+
+    stub.flush();
+    const [a, b, c] = await Promise.all([first, second, lookup]);
+    assert.equal(a, b);
+    assert.equal(a, c);
+    assert.equal(stub.calls.length, 1);
+  });
+});
+
+test('acquire rejects on ffmpeg failure, leaves no file, and can retry', async () => {
+  await withCache({ fail: true }, {}, async (cache, stub) => {
+    await assert.rejects(
+      cache.acquire(SOURCE, { startTime: 10, endTime: 14 }),
+      /FFmpeg media window failed: boom/,
+    );
+    assert.equal(cache.currentWindow, null);
+    assert.equal(await cache.lookup(SOURCE, { startTime: 10, endTime: 14 }), null);
+
+    await assert.rejects(cache.acquire(SOURCE, { startTime: 10, endTime: 14 }));
+    assert.equal(stub.calls.length, 2);
+  });
+  await withCache({ empty: true }, {}, async (cache) => {
+    await assert.rejects(
+      cache.acquire(SOURCE, { startTime: 10, endTime: 14 }),
+      /exited without creating a media window/,
+    );
+  });
+});
+
+test('acquire refuses invalid and oversized ranges without spawning ffmpeg', async () => {
+  await withCache({}, {}, async (cache, stub) => {
+    await assert.rejects(cache.acquire(SOURCE, { startTime: 10, endTime: 10 }), /invalid/);
+    await assert.rejects(cache.acquire(SOURCE, { startTime: -1, endTime: 10 }), /invalid/);
+    await assert.rejects(
+      cache.acquire(SOURCE, { startTime: 0, endTime: REMOTE_MEDIA_WINDOW_MAX_SECONDS + 1 }),
+      /too long/,
+    );
+    assert.equal(stub.calls.length, 0);
+  });
+});
+
+test('the window is deleted after the idle timeout and on cleanup', async () => {
+  await withCache({}, { idleTtlMs: 20 }, async (cache) => {
+    const window = await cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+    await new Promise((resolve) => setTimeout(resolve, 60));
+
+    assert.equal(cache.currentWindow, null);
+    assert.equal(fs.existsSync(window.path), false);
+
+    const again = await cache.acquire(SOURCE, { startTime: 10, endTime: 14 });
+    cache.cleanup();
+    assert.equal(fs.existsSync(again.path), false);
+    assert.equal(fs.existsSync(path.dirname(again.path)), false);
+  });
+});

src/core/services/remote-media-window-cache.ts

@@ -0,0 +1,377 @@
+import { execFile as nodeExecFile, type ExecFileException } from 'child_process';
+import * as fs from 'fs';
+import * as os from 'os';
+import * as path from 'path';
+import { createLogger } from '../../logger';
+import { normalizeMediaInput, type MediaInput, type MediaInputOptions } from '../../media-input';
+
+const log = createLogger('media-window');
+
+export const REMOTE_MEDIA_WINDOW_TIMEOUT_MS = 120_000;
+export const REMOTE_MEDIA_WINDOW_MAX_SECONDS = 180;
+const HEAD_SLACK_SECONDS = 0.25;
+const TAIL_SLACK_SECONDS = 1;
+const DEFAULT_IDLE_TTL_MS = 10 * 60_000;
+const COVERAGE_EPSILON_SECONDS = 0.01;
+
+export interface RemoteMediaWindowSource {
+  path: string;
+  inputOptions?: MediaInputOptions;
+  /** FFmpeg stream index to keep; `null`/undefined keeps every audio stream. */
+  audioStreamIndex?: number | null;
+}
+
+export interface RemoteMediaWindowRange {
+  startTime: number;
+  endTime: number;
+}
+
+export interface RemoteMediaWindow {
+  path: string;
+  startTime: number;
+  endTime: number;
+  sourcePath: string;
+  audioStreamIndex: number | null;
+  /** Input descriptor for FFmpeg reads; timestamps stay absolute so callers keep source times. */
+  media: MediaInput;
+}
+
+type WindowExecFile = (
+  file: string,
+  args: readonly string[],
+  options: { timeout: number },
+  callback: (error: ExecFileException | null) => void,
+) => void;
+
+export interface RemoteMediaWindowCacheOptions {
+  tempDir?: string;
+  execFile?: WindowExecFile;
+  idleTtlMs?: number;
+  logDebug?: (message: string) => void;
+}
+
+interface PendingFetch extends RemoteMediaWindowRange {
+  sourcePath: string;
+  audioStreamIndex: number | null;
+  promise: Promise<RemoteMediaWindow>;
+}
+
+export function isRemoteMediaWindowSourcePath(value: string): boolean {
+  return /^https?:\/\//i.test(value.trim());
+}
+
+function describeSourceForDebugLog(sourcePath: string): string {
+  try {
+    return `remote:${new URL(sourcePath).hostname.toLowerCase() || 'unknown'}`;
+  } catch {
+    return 'remote:unknown';
+  }
+}
+
+function isUsableRange(range: RemoteMediaWindowRange, allowEmpty: boolean): boolean {
+  return (
+    Number.isFinite(range.startTime) &&
+    Number.isFinite(range.endTime) &&
+    range.startTime >= 0 &&
+    (allowEmpty ? range.endTime >= range.startTime : range.endTime > range.startTime)
+  );
+}
+
+function audioStreamMatches(
+  windowIndex: number | null,
+  requested: number | null | undefined,
+): boolean {
+  return requested == null || windowIndex === requested;
+}
+
+function covers(
+  candidate: RemoteMediaWindowRange & { sourcePath: string; audioStreamIndex: number | null },
+  source: RemoteMediaWindowSource,
+  range: RemoteMediaWindowRange,
+): boolean {
+  return (
+    candidate.sourcePath === source.path &&
+    audioStreamMatches(candidate.audioStreamIndex, source.audioStreamIndex) &&
+    candidate.startTime <= range.startTime + COVERAGE_EPSILON_SECONDS &&
+    candidate.endTime >= range.endTime - COVERAGE_EPSILON_SECONDS
+  );
+}
+
+/**
+ * Stream-copies `[startTime, endTime]` of a remote source into a local Matroska file.
+ * `-copyts -start_at_zero` keeps the source timestamps, so later reads seek with the
+ * original times via `-seek_timestamp 1` (see `MediaInput.absoluteTimestamps`).
+ */
+export function buildRemoteMediaWindowArgs(
+  source: RemoteMediaWindowSource,
+  range: RemoteMediaWindowRange,
+  outputPath: string,
+): string[] {
+  const input = normalizeMediaInput({ path: source.path, inputOptions: source.inputOptions });
+  const audioMap =
+    typeof source.audioStreamIndex === 'number' && Number.isInteger(source.audioStreamIndex)
+      ? `0:${source.audioStreamIndex}`
+      : '0:a';
+  return [
+    '-hide_banner',
+    '-nostdin',
+    '-loglevel',
+    'error',
+    '-ss',
+    String(range.startTime),
+    '-t',
+    String(range.endTime - range.startTime),
+    ...input.inputArgs,
+    '-i',
+    input.path,
+    '-map',
+    '0:v:0?',
+    '-map',
+    audioMap,
+    '-c',
+    'copy',
+    '-sn',
+    '-dn',
+    '-copyts',
+    '-start_at_zero',
+    '-f',
+    'matroska',
+    '-y',
+    outputPath,
+  ];
+}
+
+/**
+ * Holds one downloaded window of the current remote stream so the timing review,
+ * audio extraction, and screenshot all read the same local bytes instead of each
+ * re-fetching the clip over HTTP. A new window replaces the old one; the file is
+ * deleted after `idleTtlMs` without use, on `clear()`, or on `cleanup()`.
+ */
+export class RemoteMediaWindowCache {
+  private readonly tempDir: string;
+  private readonly execFile: WindowExecFile;
+  private readonly idleTtlMs: number;
+  private readonly logDebug: (message: string) => void;
+  private current: RemoteMediaWindow | null = null;
+  private pending: PendingFetch | null = null;
+  private idleTimer: ReturnType<typeof setTimeout> | null = null;
+  private sequence = 0;
+
+  constructor(options: RemoteMediaWindowCacheOptions = {}) {
+    this.tempDir = options.tempDir ?? path.join(os.tmpdir(), 'subminer-media-windows');
+    this.execFile = options.execFile ?? nodeExecFile;
+    this.idleTtlMs = options.idleTtlMs ?? DEFAULT_IDLE_TTL_MS;
+    this.logDebug = options.logDebug ?? ((message) => log.debug(message));
+  }
+
+  get currentWindow(): RemoteMediaWindow | null {
+    return this.current;
+  }
+
+  /** Returns a ready or in-flight window covering the range; never starts a download. */
+  async lookup(
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ): Promise<RemoteMediaWindow | null> {
+    if (!isUsableRange(range, true)) return null;
+    if (this.current && covers(this.current, source, range)) {
+      this.touch();
+      return this.current;
+    }
+    const pending = this.pending;
+    if (pending && covers(pending, source, range)) {
+      try {
+        const window = await pending.promise;
+        this.touch();
+        return window;
+      } catch {
+        return null;
+      }
+    }
+    return null;
+  }
+
+  /** Returns a window covering the range, downloading (and widening) one when needed. */
+  async acquire(
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ): Promise<RemoteMediaWindow> {
+    if (!isUsableRange(range, false)) {
+      throw new Error('Media window range is invalid.');
+    }
+    if (range.endTime - range.startTime > REMOTE_MEDIA_WINDOW_MAX_SECONDS) {
+      throw new Error('Media window range is too long to download.');
+    }
+
+    for (;;) {
+      const hit = await this.lookup(source, range);
+      if (hit) return hit;
+      const pending = this.pending;
+      if (!pending) break;
+      // Another caller is already downloading; wait for it, then re-check coverage.
+      await pending.promise.catch(() => null);
+    }
+
+    return this.fetch(source, this.planFetchRange(source, range));
+  }
+
+  clear(): void {
+    this.cancelIdleTimer();
+    const current = this.current;
+    this.current = null;
+    if (current) this.removeFile(current.path);
+  }
+
+  cleanup(): void {
+    this.clear();
+    try {
+      fs.rmSync(this.tempDir, { recursive: true, force: true });
+    } catch (error) {
+      log.error('Failed to cleanup media window directory:', error);
+    }
+  }
+
+  private planFetchRange(
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ): RemoteMediaWindowRange {
+    let startTime = Math.max(0, range.startTime - HEAD_SLACK_SECONDS);
+    let endTime = range.endTime + TAIL_SLACK_SECONDS;
+    const current = this.current;
+    if (
+      current &&
+      current.sourcePath === source.path &&
+      audioStreamMatches(current.audioStreamIndex, source.audioStreamIndex)
+    ) {
+      // Keep what was already downloaded when the review timeline grows in one direction.
+      const unionStart = Math.min(startTime, current.startTime);
+      const unionEnd = Math.max(endTime, current.endTime);
+      if (unionEnd - unionStart <= REMOTE_MEDIA_WINDOW_MAX_SECONDS) {
+        startTime = unionStart;
+        endTime = unionEnd;
+      }
+    }
+    return { startTime, endTime };
+  }
+
+  private fetch(
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ): Promise<RemoteMediaWindow> {
+    fs.mkdirSync(this.tempDir, { recursive: true });
+    this.sequence += 1;
+    const outputPath = path.join(this.tempDir, `window_${Date.now()}_${this.sequence}.mkv`);
+    const audioStreamIndex =
+      typeof source.audioStreamIndex === 'number' ? source.audioStreamIndex : null;
+    const description = describeSourceForDebugLog(source.path);
+    const startedAt = Date.now();
+    this.logDebug(
+      `[media-window] fetch start ${description} start=${range.startTime} end=${range.endTime} audioStream=${audioStreamIndex ?? 'all'}`,
+    );
+
+    const promise = new Promise<RemoteMediaWindow>((resolve, reject) => {
+      this.execFile(
+        'ffmpeg',
+        buildRemoteMediaWindowArgs(source, range, outputPath),
+        { timeout: REMOTE_MEDIA_WINDOW_TIMEOUT_MS },
+        (error) => {
+          const elapsedMs = Math.max(0, Date.now() - startedAt);
+          const size = error ? 0 : this.fileSize(outputPath);
+          if (error || size === 0) {
+            this.removeFile(outputPath);
+            const reason = error
+              ? error.code === 'ENOENT'
+                ? 'FFmpeg not found. Install FFmpeg to enable media generation.'
+                : `FFmpeg media window failed: ${error.message}`
+              : 'FFmpeg exited without creating a media window.';
+            this.logDebug(`[media-window] fetch failed ${description} elapsedMs=${elapsedMs}`);
+            reject(new Error(reason));
+            return;
+          }
+          const window: RemoteMediaWindow = {
+            path: outputPath,
+            startTime: range.startTime,
+            endTime: range.endTime,
+            sourcePath: source.path,
+            audioStreamIndex,
+            media: {
+              path: outputPath,
+              source: 'remote-window',
+              singleResolvedStream: true,
+              absoluteTimestamps: true,
+            },
+          };
+          this.logDebug(
+            `[media-window] fetch complete ${description} elapsedMs=${elapsedMs} bytes=${size}`,
+          );
+          this.replaceCurrent(window);
+          resolve(window);
+        },
+      );
+    });
+
+    const pending: PendingFetch = {
+      sourcePath: source.path,
+      audioStreamIndex,
+      startTime: range.startTime,
+      endTime: range.endTime,
+      promise,
+    };
+    this.pending = pending;
+    promise
+      .catch(() => undefined)
+      .then(() => {
+        if (this.pending === pending) this.pending = null;
+      });
+    return promise;
+  }
+
+  private replaceCurrent(window: RemoteMediaWindow): void {
+    const previous = this.current;
+    this.current = window;
+    if (previous && previous.path !== window.path) this.removeFile(previous.path);
+    this.touch();
+  }
+
+  private touch(): void {
+    this.cancelIdleTimer();
+    if (this.idleTtlMs <= 0 || !this.current) return;
+    const timer = setTimeout(() => {
+      if (this.idleTimer === timer) this.idleTimer = null;
+      this.clear();
+    }, this.idleTtlMs);
+    timer.unref?.();
+    this.idleTimer = timer;
+  }
+
+  private cancelIdleTimer(): void {
+    if (this.idleTimer) clearTimeout(this.idleTimer);
+    this.idleTimer = null;
+  }
+
+  private fileSize(filePath: string): number {
+    try {
+      return fs.statSync(filePath).size;
+    } catch {
+      return 0;
+    }
+  }
+
+  private removeFile(filePath: string): void {
+    try {
+      fs.unlinkSync(filePath);
+    } catch (error) {
+      if ((error as NodeJS.ErrnoException).code !== 'ENOENT') {
+        log.debug(`Failed to remove media window ${filePath}:`, (error as Error).message);
+      }
+    }
+  }
+}
+
+let sharedCache: RemoteMediaWindowCache | null = null;
+
+/** Process-wide cache so the review modal and card media generation share one download. */
+export function getSharedRemoteMediaWindowCache(): RemoteMediaWindowCache {
+  sharedCache ??= new RemoteMediaWindowCache();
+  return sharedCache;
+}

src/core/services/subtitle-cue-parser.test.ts

@@ -1449,15 +1449,17 @@ test('parseSubtitleCues keeps tall CC-style base dialogue publishable after remo
     'Dialogue: 0,0:00:06.11,0:00:10.11,Default,,0,0,0,,{\\pos(172,437)\\fscx50}（{\\fscx100}立希{\\fscx50}）',
     'Dialogue: 0,0:00:06.11,0:00:10.11,Default,,0,0,0,,{\\pos(332,443)\\fscx50\\fscy50}ともり',
     'Dialogue: 0,0:00:06.11,0:00:10.11,Default,,0,0,0,,{\\pos(192,497)}お前…{\\fscx50}　{\\fscx100}燈をバンドに誘ったの？',
+    // A second labeled turn, so the script reads as broadcast captions.
+    'Dialogue: 0,0:00:10.11,0:00:12.00,Default,,0,0,0,,{\\pos(192,497)\\fscx50}（{\\fscx100}燈{\\fscx50}）{\\fscx100}うん。',
   ].join('\n');
 
   const cues = parseSubtitleCues(content, 'test.ass');
+  // The bare speaker label row joins the dialogue row beneath it as one cue.
   assert.deepEqual(
     cues.map((cue) => cue.text),
-    ['（立希）', 'お前…　燈をバンドに誘ったの？'],
+    ['（立希）\nお前…　燈をバンドに誘ったの？', '（燈）うん。'],
   );
-  assert.deepEqual(cues[0]?.assFurigana, ['たき']);
-  assert.deepEqual(cues[1]?.assFurigana, ['ともり']);
+  assert.deepEqual(cues[0]?.assFurigana, ['たき', 'ともり']);
   assert.ok(cues.every((cue) => cue.assLayout?.kind === 'positioned'));
 });
 
@@ -1485,14 +1487,184 @@ test('parseSubtitleCues removes half-size positioned furigana from broadcast cap
     [
       '（山田）ごめん　結局　ぬれたな。',
       '大丈夫。',
-      '（山田の母）ほんなら',
-      '隠し貯蔵のミルクまんじゅう➡',
+      '（山田の母）ほんなら\n隠し貯蔵のミルクまんじゅう➡',
       '絶対違う',
     ],
   );
   assert.deepEqual(cues[1]?.assFurigana, ['だいじょうぶ']);
-  assert.deepEqual(cues[3]?.assFurigana, ['かく', 'ちょぞう']);
-  assert.deepEqual(cues[4]?.assFurigana, ['ぜったい　ちが']);
+  assert.deepEqual(cues[2]?.assFurigana, ['かく', 'ちょぞう']);
+  assert.deepEqual(cues[3]?.assFurigana, ['ぜったい　ちが']);
+});
+
+// Broadcast-caption rows from You and I Are Polar Opposites S02E09. Every pair shares
+// timing, style, and the bottom band; only the text tells a wrap from a second speaker.
+const captionRowsHeader = ['[Script Info]', 'PlayResY: 540', '', ...eventsHeader];
+
+function captionRow(start: string, end: string, x: number, y: number, text: string): string {
+  return `Dialogue: 0,${start},${end},Default,,0,0,0,,{\\pos(${x},${y})}${text}`;
+}
+
+test('parseSubtitleCues joins caption rows that wrap one sentence across two events', () => {
+  const content = [
+    ...captionRowsHeader,
+    captionRow('0:00:19.08', '0:00:22.66', 172, 437, '⸨ぶっちゃけ'),
+    captionRow(
+      '0:00:19.08',
+      '0:00:22.66',
+      172,
+      497,
+      '早く{\\fscx50}　{\\fscx100}この勉強生活　終えたいし⸩',
+    ),
+    // No bracket at all: the upper row simply has not reached sentence punctuation.
+    captionRow('0:02:42.33', '0:02:44.43', 232, 437, '（東）≪好きだと'),
+    captionRow('0:02:42.33', '0:02:44.43', 232, 497, '自覚してしまったものの➡'),
+    // Rows are centred independently, so a wrap can change x between rows.
+    captionRow('0:00:42.21', '0:00:45.21', 252, 407, '≪ちょっとしたことで'),
+    captionRow('0:00:42.21', '0:00:45.21', 292, 497, '勝手に落ち込んだり➡'),
+    // A quote closed with 」 inside a still-open ≪…≫ span is not the end of the line.
+    captionRow('0:19:02.84', '0:19:05.00', 212, 437, '≪「つきあえる自信がない」'),
+    captionRow('0:19:02.84', '0:19:05.00', 452, 497, 'じゃない≫'),
+    // An in-sentence 「 quote on the lower row is not a new turn.
+    captionRow('0:18:35.55', '0:18:38.00', 232, 437, '今　「好きだ」と'),
+    captionRow('0:18:35.55', '0:18:38.00', 192, 497, '「心地いい」と感じてるのも➡'),
+  ].join('\n');
+
+  const cues = parseSubtitleCues(content, 'polar-opposites-s02e09.ass');
+
+  assert.deepEqual(
+    cues.map((cue) => cue.text),
+    [
+      '⸨ぶっちゃけ\n早く　この勉強生活　終えたいし⸩',
+      '≪ちょっとしたことで\n勝手に落ち込んだり➡',
+      '（東）≪好きだと\n自覚してしまったものの➡',
+      '今　「好きだ」と\n「心地いい」と感じてるのも➡',
+      '≪「つきあえる自信がない」\nじゃない≫',
+    ],
+  );
+  assert.ok(cues.every((cue) => cue.assLayout?.kind === 'positioned'));
+});
+
+test('parseSubtitleCues keeps simultaneous caption rows from two speakers separate', () => {
+  const content = [
+    ...captionRowsHeader,
+    // Both unlabeled: the upper row finished its sentence.
+    captionRow('0:03:56.10', '0:04:00.04', 172, 437, 'なあ　車両　変えね？'),
+    captionRow('0:03:56.10', '0:04:00.04', 632, 497, 'えっ？➡'),
+    // Lower row opens a labeled turn.
+    captionRow('0:03:38.48', '0:03:42.05', 592, 437, 'おはよう！'),
+    captionRow('0:03:38.48', '0:03:42.05', 272, 497, '（平）あっ　声　でかっ。'),
+    // A closed monologue span above a sound effect.
+    captionRow('0:08:16.83', '0:08:19.50', 372, 437, '≪落ち着け　落ち着け≫'),
+    captionRow('0:08:16.83', '0:08:19.50', 272, 497, 'ドクン　ドクン　ドクン…'),
+    // Two labeled speakers.
+    captionRow('0:09:27.90', '0:09:31.07', 312, 437, '（平）ぐぅ…。'),
+    captionRow('0:09:27.90', '0:09:31.07', 352, 497, '（東）≪ちくしょう～！≫'),
+    // A bare label never swallows a differently labeled row.
+    captionRow('0:11:43.24', '0:11:45.00', 212, 437, '（長谷川）'),
+    captionRow('0:11:43.24', '0:11:45.00', 412, 497, '（早乙女）ん？'),
+    // A short sentence-final 。 closes the upper row like any other.
+    captionRow('0:12:31.55', '0:12:33.55', 172, 437, '⚞（東）平。'),
+    captionRow('0:12:31.55', '0:12:33.55', 532, 497, 'あっ。'),
+  ].join('\n');
+
+  const cues = parseSubtitleCues(content, 'polar-opposites-s02e09.ass');
+
+  assert.deepEqual(
+    cues.map((cue) => cue.text),
+    [
+      'おはよう！',
+      '（平）あっ　声　でかっ。',
+      'なあ　車両　変えね？',
+      'えっ？➡',
+      '≪落ち着け　落ち着け≫',
+      'ドクン　ドクン　ドクン…',
+      '（平）ぐぅ…。',
+      '（東）≪ちくしょう～！≫',
+      '（長谷川）',
+      '（早乙女）ん？',
+      '⚞（東）平。',
+      'あっ。',
+    ],
+  );
+});
+
+test('parseSubtitleCues keeps caption rows apart across styles, bands, and timing', () => {
+  const content = [
+    ...captionRowsHeader,
+    // Same wording as a wrap, but the rows sit in different vertical bands.
+    captionRow('0:01:00.00', '0:01:02.00', 172, 77, '≪ちょっとしたことで'),
+    captionRow('0:01:00.00', '0:01:02.00', 172, 497, '勝手に落ち込んだり➡'),
+    // Same band, but a sign style beside dialogue.
+    'Dialogue: 0,0:01:05.00,0:01:07.00,Sign,,0,0,0,,{\\pos(172,437)}ちょっとしたことで',
+    captionRow('0:01:05.00', '0:01:07.00', 172, 497, '勝手に落ち込んだり➡'),
+    // Same rows, but the lower one ends later.
+    captionRow('0:01:10.00', '0:01:12.00', 172, 437, '≪ちょっとしたことで'),
+    captionRow('0:01:10.00', '0:01:13.00', 172, 497, '勝手に落ち込んだり➡'),
+    // Style-aligned rows without \pos are never caption rows.
+    'Dialogue: 0,0:01:15.00,0:01:17.00,Default,,0,0,0,,{\\an8}≪ちょっとしたことで',
+    'Dialogue: 0,0:01:15.00,0:01:17.00,Default,,0,0,0,,{\\an2}勝手に落ち込んだり➡',
+    // Same height: the events sit side by side, not one above the other.
+    captionRow('0:01:20.00', '0:01:22.00', 172, 497, '≪ちょっとしたことで'),
+    captionRow('0:01:20.00', '0:01:22.00', 612, 497, '勝手に落ち込んだり➡'),
+    // Same bottom band, but further apart than two text rows.
+    captionRow('0:01:25.00', '0:01:27.00', 172, 367, '≪ちょっとしたことで'),
+    captionRow('0:01:25.00', '0:01:27.00', 172, 497, '勝手に落ち込んだり➡'),
+  ].join('\n');
+
+  const cues = parseSubtitleCues(content, 'test.ass');
+
+  assert.equal(cues.length, 12);
+  assert.ok(cues.every((cue) => !cue.text.includes('\n')));
+});
+
+test('parseSubtitleCues leaves typeset rows alone in scripts that are not broadcast captions', () => {
+  // Fansub typesetting stacks positioned rows for signs, chat bubbles, and headlines. Such
+  // text carries no caption punctuation, so without the script-level gate every stacked
+  // pair here would read as an unfinished sentence and merge.
+  const content = [
+    ...captionRowsHeader,
+    captionRow('0:00:10.00', '0:00:14.00', 640, 200, 'Shocking Statement Leaves'),
+    captionRow('0:00:10.00', '0:00:14.00', 640, 260, 'Listeners Speechless!'),
+    captionRow('0:01:00.00', '0:01:04.00', 400, 300, 'shes here AGAIN'),
+    captionRow('0:01:00.00', '0:01:04.00', 400, 360, 'make sakiko-chan go home'),
+    // Japanese typesetting in the same script is held back by the same gate.
+    captionRow('0:02:00.00', '0:02:04.00', 300, 400, '定休日'),
+    captionRow('0:02:00.00', '0:02:04.00', 300, 460, '毎週水曜日'),
+  ].join('\n');
+
+  const cues = parseSubtitleCues(content, 'test.ass');
+
+  assert.deepEqual(
+    cues.map((cue) => cue.text),
+    [
+      'Shocking Statement Leaves',
+      'Listeners Speechless!',
+      'shes here AGAIN',
+      'make sakiko-chan go home',
+      '定休日',
+      '毎週水曜日',
+    ],
+  );
+});
+
+test('parseSubtitleCues never joins caption rows that carry no Japanese', () => {
+  // Even inside a caption script, romaji or English rows are not the wrapped Japanese
+  // sentences this pass targets.
+  const content = [
+    ...captionRowsHeader,
+    captionRow('0:00:10.00', '0:00:13.00', 172, 437, '（東）≪好きだと'),
+    captionRow('0:00:10.00', '0:00:13.00', 172, 497, '自覚してしまったものの➡'),
+    captionRow('0:00:20.00', '0:00:23.00', 172, 437, '（平）ん？'),
+    captionRow('0:00:30.00', '0:00:34.00', 640, 437, 'NOW LOADING'),
+    captionRow('0:00:30.00', '0:00:34.00', 640, 497, 'please wait'),
+  ].join('\n');
+
+  const cues = parseSubtitleCues(content, 'test.ass');
+
+  assert.deepEqual(
+    cues.map((cue) => cue.text),
+    ['（東）≪好きだと\n自覚してしまったものの➡', '（平）ん？', 'NOW LOADING', 'please wait'],
+  );
 });
 
 test('parseSubtitleCues scales furigana geometry by PlayResY', () => {

src/core/services/subtitle-cue-parser.ts

@@ -2342,6 +2342,185 @@ function removeAssFuriganaEvents(
   };
 }
 
+// Broadcast-caption converters give every visual row of one utterance its own positioned
+// event, so a sentence that wraps arrives as two simultaneous cues with the same timing,
+// style, and vertical band. Captions punctuate every finished utterance, and each turn
+// opens with a speaker label or a ≪…≫ / ⸨…⸩ span, which is what tells a wrapped sentence
+// apart from two speakers sharing the screen.
+const CAPTION_SPEAKER_LABEL_ONLY_PATTERN = /^（[^（）]*）$/u;
+const CAPTION_SPEAKER_LABEL_PATTERN = /^（/u;
+const CAPTION_TURN_OPENER_PATTERN = /^[≪⸨（]/u;
+const CAPTION_TERMINAL_PATTERN = /[。？！?!…‥～〜➡⁉⁈≫⸩）」』]$/u;
+const CAPTION_SPANS: ReadonlyArray<readonly [open: string, close: string]> = [
+  ['≪', '≫'],
+  ['⸨', '⸩'],
+];
+// Rows of one utterance sit one text row apart (about 60 units in the 540-line space the
+// furigana geometry is tuned for), or two when a ruby row lies between them. Rows at the
+// same height sit side by side, and rows further apart are separate placements.
+const MAX_CAPTION_ROW_GAP = 120;
+// Only a broadcast-caption script gets rows joined. Typesetters position rows for signs,
+// chat bubbles, and lyric stacks too, and there the continuation rule below has no
+// convention to read: sign text rarely carries sentence punctuation, so unrelated rows
+// would run together. A caption script announces itself by labelling speakers （名） and
+// bracketing off-screen speech in ≪…≫ / ⸨…⸩; typeset scripts use those in a handful of
+// lines at most. Measured over local tracks, caption scripts sit near 25% and every typeset
+// script below 1%, so the threshold has room on both sides. It is deliberately strict: a
+// caption script wrongly held back just keeps one sentence on two rows, while a typeset
+// script wrongly let through concatenates unrelated signs.
+const MIN_CAPTION_EVIDENCE_EVENTS = 2;
+const MIN_CAPTION_EVIDENCE_RATIO = 0.05;
+const CAPTION_EVIDENCE_PATTERN = /^（[^（）]{1,14}）|[≪⸨]/u;
+// Rows that carry no Japanese are not the broadcast captions this pass targets.
+const JAPANESE_SCRIPT_PATTERN = /[\p{Script=Hiragana}\p{Script=Katakana}\p{Script=Han}]/u;
+
+function hasBroadcastCaptionConventions(cues: readonly AnnotatedSubtitleCue[]): boolean {
+  let evidence = 0;
+  let published = 0;
+  for (const cue of cues) {
+    if (!cue.text.trim()) continue;
+    published += 1;
+    if (CAPTION_EVIDENCE_PATTERN.test(cue.text)) evidence += 1;
+  }
+  return (
+    evidence >= MIN_CAPTION_EVIDENCE_EVENTS && evidence >= published * MIN_CAPTION_EVIDENCE_RATIO
+  );
+}
+
+function captionSpanDepth(text: string, [open, close]: readonly [string, string]): number {
+  let depth = 0;
+  for (const char of text) {
+    if (char === open) depth += 1;
+    else if (char === close) depth -= 1;
+  }
+  return depth;
+}
+
+/**
+ * Whether `lower` continues the utterance `upper` started, both being simultaneous
+ * caption rows. A bare speaker label labels the row beneath it. Otherwise the upper row
+ * must not have finished: it ends without terminal punctuation, or a ≪…≫ / ⸨…⸩ span it
+ * opened is still open (closing 」 inside such a span is not an ending). A lower row that
+ * opens its own turn is always a different line.
+ */
+function isCaptionRowContinuation(upper: string, lower: string): boolean {
+  if (CAPTION_SPEAKER_LABEL_ONLY_PATTERN.test(upper)) {
+    return !CAPTION_SPEAKER_LABEL_PATTERN.test(lower);
+  }
+  if (CAPTION_TURN_OPENER_PATTERN.test(lower)) {
+    return false;
+  }
+  const spanContinues = CAPTION_SPANS.some(
+    (span) => captionSpanDepth(upper, span) > 0 || captionSpanDepth(lower, span) < 0,
+  );
+  return spanContinues || !CAPTION_TERMINAL_PATTERN.test(upper);
+}
+
+// A half-height row is ruby or a whispered aside, not a row of the utterance.
+function isCaptionRowCandidate(cue: AnnotatedSubtitleCue): boolean {
+  const scaleY = staticAssScalePercent(cue, 'fscy');
+  return (
+    cue.source === undefined &&
+    cue.assLayout?.kind === 'positioned' &&
+    cue.effect.trim() === '' &&
+    !cue.text.includes('\n') &&
+    !hasAssTemporalOverride(cue.overrides) &&
+    (scaleY === null || scaleY > MAX_ASS_FURIGANA_SCALE_PERCENT) &&
+    JAPANESE_SCRIPT_PATTERN.test(cue.text)
+  );
+}
+
+function captionRowGroupKey(cue: AnnotatedSubtitleCue): string {
+  return [
+    cue.startTime,
+    cue.endTime,
+    cue.style,
+    cue.layer,
+    cue.name,
+    cue.assLayout?.verticalBand ?? '',
+  ].join('\0');
+}
+
+function mergeCaptionRows(rows: readonly AnnotatedSubtitleCue[]): AnnotatedSubtitleCue {
+  const [first] = rows;
+  if (!first) throw new Error('mergeCaptionRows requires at least one row');
+  const overrides = rows.flatMap((row) => row.overrides);
+  const assFurigana = [...new Set(rows.flatMap((row) => row.assFurigana ?? []))];
+  return {
+    ...first,
+    text: rows.map((row) => row.text).join('\n'),
+    rawText: rows.map((row) => row.rawText).join('\\N'),
+    overrides,
+    overrideSignature: assOverrideSignature(overrides),
+    ...(assFurigana.length === 0 ? {} : { assFurigana }),
+  };
+}
+
+/**
+ * Join simultaneous caption rows that spell one utterance into a single cue, so the
+ * overlay can wrap or flatten it like an authored `\N` line and the sidebar and mining
+ * paths see the whole sentence. Rows stack top to bottom; each row joins the cue above it
+ * only while `isCaptionRowContinuation` holds, so a second speaker starts a new cue. The
+ * whole pass is skipped unless the script reads as broadcast captions.
+ */
+function mergeAssCaptionRows(
+  cues: AnnotatedSubtitleCue[],
+  playResY: number | null,
+): AnnotatedSubtitleCue[] {
+  if (!hasBroadcastCaptionConventions(cues)) return cues;
+
+  const groups = new Map<string, AnnotatedSubtitleCue[]>();
+  for (const cue of cues) {
+    if (!isCaptionRowCandidate(cue)) continue;
+    const key = captionRowGroupKey(cue);
+    const group = groups.get(key);
+    if (group) group.push(cue);
+    else groups.set(key, [cue]);
+  }
+
+  const maxRowGap = MAX_CAPTION_ROW_GAP * assFuriganaGeometryScale(playResY);
+  const rowY = (cue: AnnotatedSubtitleCue): number =>
+    cue.assLayout?.kind === 'positioned' ? cue.assLayout.y : 0;
+  const replacements = new Map<AnnotatedSubtitleCue, AnnotatedSubtitleCue>();
+  const removed = new Set<AnnotatedSubtitleCue>();
+  for (const group of groups.values()) {
+    if (group.length < 2) continue;
+    const rows = [...group].sort((a, b) => rowY(a) - rowY(b) || a.order - b.order);
+    let run: AnnotatedSubtitleCue[] = [];
+    const flush = (): void => {
+      if (run.length < 2) return;
+      const anchor = run.reduce((lowest, row) => (row.order < lowest.order ? row : lowest));
+      replacements.set(anchor, mergeCaptionRows(run));
+      for (const row of run) {
+        if (row !== anchor) removed.add(row);
+      }
+    };
+    for (const row of rows) {
+      const previous = run.at(-1);
+      const gap = previous ? rowY(row) - rowY(previous) : 0;
+      if (
+        previous &&
+        gap > 0 &&
+        gap <= maxRowGap &&
+        previous.text !== row.text &&
+        isCaptionRowContinuation(previous.text, row.text)
+      ) {
+        run.push(row);
+        continue;
+      }
+      flush();
+      run = [row];
+    }
+    flush();
+  }
+
+  if (replacements.size === 0) return cues;
+  return cues.flatMap((cue) => {
+    if (removed.has(cue)) return [];
+    return [replacements.get(cue) ?? cue];
+  });
+}
+
 function parseAnnotatedAssEvents(content: string, placement: AssPlacementContext): ParsedAssEvents {
   const cues: AnnotatedSubtitleCue[] = [];
   const comments: AnnotatedSubtitleCue[] = [];
@@ -2476,7 +2655,10 @@ function parseAnnotatedAssCues(content: string): AnnotatedSubtitleCue[] {
     removeAssFontTextureEvents(parseAnnotatedAssEvents(content, placement)),
     placement.playResY,
   );
-  return recoverFragmentOnlyAssLines(recoverCanonicalAssEvents(events));
+  return mergeAssCaptionRows(
+    recoverFragmentOnlyAssLines(recoverCanonicalAssEvents(events)),
+    placement.playResY,
+  );
 }
 
 export function parseAssCues(content: string): SubtitleCue[] {

src/core/services/subtitle-timing-offset.test.ts

@@ -1,73 +0,0 @@
-import assert from 'node:assert/strict';
-import test from 'node:test';
-import { estimateSubtitleTimingOffset } from './subtitle-timing-offset';
-
-function cue(startTime: number) {
-  return { startTime, endTime: startTime + 1, text: `cue ${startTime}` };
-}
-
-test('estimate subtitle timing offset detects a late Jellyfin subtitle timeline', () => {
-  const primary = [
-    34.935, 36.937, 41.441, 45.279, 48.115, 52.286, 54.955, 59.793, 63.63, 67.634, 76.643, 80.814,
-    87.988, 90.991, 94.094, 97.097,
-  ].map(cue);
-  const reference = [
-    3.46, 9.48, 13.61, 21.4, 28.16, 32.06, 35.93, 45.1, 56.57, 59.68, 62.44, 65.56,
-  ].map(cue);
-
-  const result = estimateSubtitleTimingOffset(primary, reference);
-
-  assert.ok(result);
-  assert.ok(result.offsetSeconds > -32);
-  assert.ok(result.offsetSeconds < -31);
-  assert.ok(result.matchCount >= 8);
-  assert.ok(result.meanErrorSeconds <= 0.75);
-});
-
-test('estimate subtitle timing offset favors the early episode timeline', () => {
-  const primary = [
-    34.935, 36.937, 41.441, 45.279, 48.115, 52.286, 54.955, 59.793, 63.63, 67.634, 76.643, 80.814,
-    87.988, 90.991, 94.094, 97.097, 207.974, 212.579, 222.422, 228.095, 232.432, 238.271, 244.778,
-    246.78, 249.282, 251.284, 253.62, 256.289, 259.626, 262.129, 264.965, 267.634, 270.303, 274.407,
-    277.077, 280.08, 284.084, 288.421, 291.925, 295.262, 298.431, 301.101, 306.773, 308.942,
-    312.946, 316.283, 321.621, 326.626, 331.131, 336.069, 340.407, 343.41, 351.418, 355.422,
-    357.924, 362.429, 365.432, 370.604, 373.273, 377.944, 381.114, 384.618, 387.621, 390.957,
-    396.73, 399.232, 401.568, 403.57, 405.572, 407.574, 409.743, 412.746, 418.752, 425.258, 427.26,
-    435.602, 440.44, 442.942, 445.445, 449.783,
-  ].map(cue);
-  const reference = [
-    3.46, 9.48, 13.61, 21.4, 28.16, 32.06, 35.93, 45.1, 56.57, 59.68, 62.44, 65.56, 165.77, 172.81,
-    176.1, 177.27, 186.33, 191.33, 195.78, 201.83, 212.9, 214.09, 216.73, 220.2, 222.91, 225.65,
-    232.8, 237.92, 242.23, 243.28, 247.53, 252.04, 255.9, 258.86, 262.09, 264.43, 276.07, 278.01,
-    280.98, 285.67, 289.89, 294.57, 300, 303.56, 308.58, 316.37, 318.38, 319.86, 325.38, 328.82,
-    333.68, 335.26, 336.82, 340.11, 342.11, 344.36, 346.39, 347.53, 350.92, 370.18, 372.88, 376.43,
-    388.2, 390.57, 403.96, 406.36, 409.72, 413.78, 425.55, 432.76, 435.03, 438.06, 443.73, 448.31,
-    450.57, 457.62, 463.41, 465.85, 473.79, 480.59,
-  ].map(cue);
-
-  const result = estimateSubtitleTimingOffset(primary, reference);
-
-  assert.ok(result);
-  assert.ok(result.offsetSeconds > -32);
-  assert.ok(result.offsetSeconds < -31);
-});
-
-test('estimate subtitle timing offset ignores subtitle timelines that are already aligned', () => {
-  const starts = [1, 5, 9, 14, 20, 25, 31, 38];
-
-  const result = estimateSubtitleTimingOffset(
-    starts.map(cue),
-    starts.map((start) => cue(start + 0.04)),
-  );
-
-  assert.equal(result, null);
-});
-
-test('estimate subtitle timing offset rejects weak timeline matches', () => {
-  const primary = [10, 20, 30, 40, 50, 60, 70, 80].map(cue);
-  const reference = [1, 2, 3, 4, 5, 6, 7, 8].map(cue);
-
-  const result = estimateSubtitleTimingOffset(primary, reference);
-
-  assert.equal(result, null);
-});

src/core/services/subtitle-timing-offset.ts

@@ -1,153 +0,0 @@
-import type { SubtitleCue } from './subtitle-cue-parser';
-
-export type SubtitleTimingOffsetResult = {
-  offsetSeconds: number;
-  matchCount: number;
-  meanErrorSeconds: number;
-  maxErrorSeconds: number;
-};
-
-export type SubtitleTimingOffsetOptions = {
-  maxCueCount?: number;
-  maxOffsetSeconds?: number;
-  matchThresholdSeconds?: number;
-  maxMeanErrorSeconds?: number;
-  minMatchCount?: number;
-  minMatchRatio?: number;
-  minUsefulOffsetSeconds?: number;
-};
-
-type OffsetScore = SubtitleTimingOffsetResult;
-
-const DEFAULT_MAX_CUE_COUNT = 60;
-const DEFAULT_MAX_OFFSET_SECONDS = 180;
-const DEFAULT_MATCH_THRESHOLD_SECONDS = 1;
-const DEFAULT_MAX_MEAN_ERROR_SECONDS = 0.75;
-const DEFAULT_MIN_MATCH_COUNT = 8;
-const DEFAULT_MIN_MATCH_RATIO = 0.25;
-const DEFAULT_MIN_USEFUL_OFFSET_SECONDS = 0.25;
-
-function normalizeCueStarts(cues: SubtitleCue[], maxCueCount: number): number[] {
-  const starts = cues
-    .map((cue) => cue.startTime)
-    .filter((start) => Number.isFinite(start) && start >= 0)
-    .sort((a, b) => a - b);
-  const deduped: number[] = [];
-  for (const start of starts) {
-    const previous = deduped[deduped.length - 1];
-    if (previous === undefined || Math.abs(start - previous) > 0.05) {
-      deduped.push(start);
-    }
-    if (deduped.length >= maxCueCount) {
-      break;
-    }
-  }
-  return deduped;
-}
-
-function roundToMillis(value: number): number {
-  return Math.round(value * 1000) / 1000;
-}
-
-function scoreOffset(
-  primaryStarts: number[],
-  referenceStarts: number[],
-  offsetSeconds: number,
-  matchThresholdSeconds: number,
-): OffsetScore {
-  let primaryIndex = 0;
-  let referenceIndex = 0;
-  let matchCount = 0;
-  let totalErrorSeconds = 0;
-  let maxErrorSeconds = 0;
-
-  while (primaryIndex < primaryStarts.length && referenceIndex < referenceStarts.length) {
-    const shiftedPrimary = primaryStarts[primaryIndex]! + offsetSeconds;
-    const reference = referenceStarts[referenceIndex]!;
-    const errorSeconds = Math.abs(shiftedPrimary - reference);
-    if (errorSeconds <= matchThresholdSeconds) {
-      matchCount += 1;
-      totalErrorSeconds += errorSeconds;
-      maxErrorSeconds = Math.max(maxErrorSeconds, errorSeconds);
-      primaryIndex += 1;
-      referenceIndex += 1;
-      continue;
-    }
-
-    if (shiftedPrimary < reference) {
-      primaryIndex += 1;
-    } else {
-      referenceIndex += 1;
-    }
-  }
-
-  return {
-    offsetSeconds,
-    matchCount,
-    meanErrorSeconds: matchCount > 0 ? totalErrorSeconds / matchCount : Number.POSITIVE_INFINITY,
-    maxErrorSeconds,
-  };
-}
-
-function isBetterScore(next: OffsetScore, current: OffsetScore | null): boolean {
-  if (current === null) return true;
-  if (next.matchCount !== current.matchCount) return next.matchCount > current.matchCount;
-  if (next.meanErrorSeconds !== current.meanErrorSeconds) {
-    return next.meanErrorSeconds < current.meanErrorSeconds;
-  }
-  return Math.abs(next.offsetSeconds) < Math.abs(current.offsetSeconds);
-}
-
-export function estimateSubtitleTimingOffset(
-  primaryCues: SubtitleCue[],
-  referenceCues: SubtitleCue[],
-  options: SubtitleTimingOffsetOptions = {},
-): SubtitleTimingOffsetResult | null {
-  const maxCueCount = options.maxCueCount ?? DEFAULT_MAX_CUE_COUNT;
-  const maxOffsetSeconds = options.maxOffsetSeconds ?? DEFAULT_MAX_OFFSET_SECONDS;
-  const matchThresholdSeconds = options.matchThresholdSeconds ?? DEFAULT_MATCH_THRESHOLD_SECONDS;
-  const maxMeanErrorSeconds = options.maxMeanErrorSeconds ?? DEFAULT_MAX_MEAN_ERROR_SECONDS;
-  const minMatchCount = options.minMatchCount ?? DEFAULT_MIN_MATCH_COUNT;
-  const minMatchRatio = options.minMatchRatio ?? DEFAULT_MIN_MATCH_RATIO;
-  const minUsefulOffsetSeconds =
-    options.minUsefulOffsetSeconds ?? DEFAULT_MIN_USEFUL_OFFSET_SECONDS;
-
-  const primaryStarts = normalizeCueStarts(primaryCues, maxCueCount);
-  const referenceStarts = normalizeCueStarts(referenceCues, maxCueCount);
-  const comparableCueCount = Math.min(primaryStarts.length, referenceStarts.length);
-  if (comparableCueCount < minMatchCount) {
-    return null;
-  }
-
-  const candidates = new Set<number>();
-  for (const primaryStart of primaryStarts) {
-    for (const referenceStart of referenceStarts) {
-      const offsetSeconds = roundToMillis(referenceStart - primaryStart);
-      if (Math.abs(offsetSeconds) <= maxOffsetSeconds) {
-        candidates.add(offsetSeconds);
-      }
-    }
-  }
-
-  let best: OffsetScore | null = null;
-  for (const offsetSeconds of candidates) {
-    if (Math.abs(offsetSeconds) < minUsefulOffsetSeconds) {
-      continue;
-    }
-    const score = scoreOffset(primaryStarts, referenceStarts, offsetSeconds, matchThresholdSeconds);
-    if (score.matchCount < minMatchCount) {
-      continue;
-    }
-    if (score.matchCount / comparableCueCount < minMatchRatio) {
-      continue;
-    }
-    if (score.meanErrorSeconds > maxMeanErrorSeconds) {
-      continue;
-    }
-    if (isBetterScore(score, best)) {
-      best = score;
-    }
-  }
-
-  return best;
-}

src/main.ts

@@ -235,7 +235,10 @@ import {
   createCycleSecondarySubModeRuntimeHandler,
 } from './main/runtime/domains/mpv';
 import { buildSubtitleTrackDiagnostics } from './main/runtime/mpv-track-diagnostics';
-import { resolveCanonicalPrimarySubtitle } from './main/runtime/primary-subtitle-text';
+import {
+  resolveCanonicalPrimarySubtitle,
+  resolvePrimarySubtitle,
+} from './main/runtime/primary-subtitle-text';
 import {
   createBuildCopyCurrentSubtitleMainDepsHandler,
   createBuildHandleMineSentenceDigitMainDepsHandler,
@@ -302,7 +305,6 @@ import {
   listJellyfinItemsRuntime,
   listJellyfinLibrariesRuntime,
   listJellyfinSubtitleTracksRuntime,
-  loadJellyfinSubtitleDelay,
   loadSubtitlePosition as loadSubtitlePositionCore,
   loadYomitanExtension as loadYomitanExtensionCore,
   markLastCardAsAudioCard as markLastCardAsAudioCardCore,
@@ -315,7 +317,6 @@ import {
   resolveSanitizedSubtitleSeekCommand,
   resolveJellyfinPlaybackPlanRuntime,
   runStartupBootstrapRuntime,
-  saveJellyfinSubtitleDelay,
   saveSubtitlePosition as saveSubtitlePositionCore,
   clearYomitanParserCachesForWindow,
   getYomitanCurrentAnkiDeckName as getYomitanCurrentAnkiDeckNameCore,
@@ -466,6 +467,8 @@ import { handleCliCommandRuntimeServiceWithContext } from './main/cli-runtime';
 import { createOverlayModalRuntimeService } from './main/overlay-runtime';
 import { createOverlayModalInputState } from './main/runtime/overlay-modal-input-state';
 import { MediaTimingPreviewSession } from './core/services/media-timing-preview';
+import { getSharedRemoteMediaWindowCache } from './core/services/remote-media-window-cache';
+import { resolveMediaGenerationInput } from './anki-integration/media-source';
 import { generateSpeechWaveform } from './core/services/media-timing-waveform';
 import {
   collectMediaTimingContextLines,
@@ -684,7 +687,6 @@ function spawnManagedMpvProcess(args: string[]): ReturnType<typeof spawn> {
 }
 
 let activeJellyfinRemotePlayback: ActiveJellyfinRemotePlaybackState | null = null;
-let activeJellyfinSubtitleDelayKey: { itemId: string; streamIndex: number } | null = null;
 let jellyfinRemoteLastProgressAtMs = 0;
 let jellyfinMpvAutoLaunchInFlight: Promise<boolean> | null = null;
 let backgroundWarmupsStarted = false;
@@ -1830,28 +1832,31 @@ function withCurrentSubtitleTiming(payload: SubtitleData): SubtitleData {
 }
 
 function captureCurrentPrimarySubtitleMiningContext(): SubtitleMiningContext | null {
-  const canonical = resolveCanonicalPrimarySubtitle({
+  // Mine what the overlay shows, not raw mpv `sub-text`: the raw text lists every active
+  // event, so a finished caption row lingering beside a fresh line would end up on the
+  // card. The parsed view also carries the cue's own timings for the clip range.
+  const resolved = resolvePrimarySubtitle({
     liveText: appState.mpvClient?.currentSubText ?? '',
     currentTimeSec: Number(appState.mpvClient?.currentTimePos),
     cues: appState.activeParsedSubtitleCues,
   });
-  // Same validity bar as the live capture path: an unusable canonical span must fall
+  // Same validity bar as the live capture path: an unusable resolved span must fall
   // back rather than hand mining an empty line or an inverted range.
-  const canonicalText = canonical?.text.trim();
+  const resolvedText = resolved?.text.replace(/\n{2,}/g, '\n').trim();
   if (
-    !canonical ||
-    !canonicalText ||
-    !Number.isFinite(canonical.startTime) ||
-    !Number.isFinite(canonical.endTime) ||
-    canonical.endTime <= canonical.startTime
+    !resolved ||
+    !resolvedText ||
+    !Number.isFinite(resolved.startTime) ||
+    !Number.isFinite(resolved.endTime) ||
+    resolved.endTime <= resolved.startTime
   ) {
     return captureLiveSubtitleMiningContext(appState.mpvClient);
   }
   return {
     source: 'overlay',
-    text: canonicalText,
-    startTime: canonical.startTime,
-    endTime: canonical.endTime,
+    text: resolvedText,
+    startTime: resolved.startTime,
+    endTime: resolved.endTime,
     capturedAtMs: Date.now(),
   };
 }
@@ -2489,7 +2494,6 @@ const fieldGroupingOverlayRuntime = createFieldGroupingOverlayRuntime<OverlayHos
 const createFieldGroupingCallback = fieldGroupingOverlayRuntime.createFieldGroupingCallback;
 
 const SUBTITLE_POSITIONS_DIR = path.join(CONFIG_DIR, 'subtitle-positions');
-const JELLYFIN_SUBTITLE_DELAYS_PATH = path.join(CONFIG_DIR, 'jellyfin-subtitle-delays.json');
 
 const mediaRuntime = createMediaRuntimeService(
   createBuildMediaRuntimeMainDepsHandler({
@@ -2900,6 +2904,21 @@ const mediaTimingReviewRuntime = createMediaTimingReviewRuntime({
     configService.getConfig().mpv.executablePath || process.env.SUBMINER_MPV_PATH?.trim() || '',
   createPreviewSession: () => new MediaTimingPreviewSession(),
   generateWaveform: (options) => generateSpeechWaveform(options),
+  resolveMediaSource: async () => {
+    const resolved = await resolveMediaGenerationInput(appState.mpvClient, 'audio', {
+      getCachedMediaPath: (currentVideoPath, kind) =>
+        getCachedYoutubeMediaPathForCurrentPlayback(currentVideoPath, kind),
+      remoteCacheMode: shouldRequireYoutubeMediaCacheForCurrentPlayback() ? 'required' : 'optional',
+    });
+    return resolved
+      ? {
+          path: resolved.path,
+          ...(resolved.inputOptions ? { inputOptions: resolved.inputOptions } : {}),
+          singleResolvedStream: resolved.singleResolvedStream,
+        }
+      : null;
+  },
+  acquireMediaWindow: (source, range) => getSharedRemoteMediaWindowCache().acquire(source, range),
   getSubtitleContextLines: (range) =>
     collectMediaTimingContextLines({
       cues: appState.activeParsedSubtitleCues,
@@ -2908,6 +2927,14 @@ const mediaTimingReviewRuntime = createMediaTimingReviewRuntime({
       endTime: range.endTime,
     }),
   openModal: (payload) => openMediaTimingReviewModal(createOverlayHostedModalOpenDeps(), payload),
+  onPreviewEnded: (reviewId) => {
+    // The review may live in either overlay window; the renderer ignores foreign review ids.
+    for (const window of [overlayManager.getMainWindow(), overlayManager.getModalWindow()]) {
+      if (window && !window.isDestroyed()) {
+        window.webContents.send(IPC_CHANNELS.event.mediaTimingReviewPreviewEnded, reviewId);
+      }
+    }
+  },
   showStatus: (message) =>
     overlayNotificationsRuntime.showConfiguredStatusNotification(message, { variant: 'warning' }),
 });
@@ -3133,23 +3160,6 @@ const {
     wait: (ms) => new Promise<void>((resolve) => setTimeout(resolve, ms)),
     cacheSubtitleTrack: (track) => jellyfinSubtitleCacheIo.cacheSubtitleTrack(track),
     cleanupCachedSubtitles: (dirs) => jellyfinSubtitleCacheIo.cleanupCachedSubtitles(dirs),
-    getSavedSubtitleDelay: (itemId, streamIndex) =>
-      loadJellyfinSubtitleDelay({
-        filePath: JELLYFIN_SUBTITLE_DELAYS_PATH,
-        itemId,
-        streamIndex,
-      }),
-    setActiveSubtitleDelayKey: (key) => {
-      activeJellyfinSubtitleDelayKey = key;
-    },
-    loadSubtitleSourceText,
-    saveSubtitleDelay: (itemId, streamIndex, delaySeconds) =>
-      saveJellyfinSubtitleDelay({
-        filePath: JELLYFIN_SUBTITLE_DELAYS_PATH,
-        itemId,
-        streamIndex,
-        delaySeconds,
-      }),
     initSubtitlePrefetch: (sourcePath) =>
       subtitlePrefetchRuntime.refreshSubtitleSidebarFromSource(sourcePath),
     logDebug: (message, error) => {
@@ -3215,7 +3225,6 @@ const {
     getActivePlayback: () => activeJellyfinRemotePlayback,
     clearActivePlayback: () => {
       activeJellyfinRemotePlayback = null;
-      activeJellyfinSubtitleDelayKey = null;
     },
     getSession: () => appState.jellyfinRemoteSession,
     getNow: () => Date.now(),
@@ -3995,6 +4004,7 @@ const {
     cleanupInternalSubtitleTrackCache: () => cachedInternalSubtitleTrackExtractor.clear(),
     cleanupYoutubeSubtitleTempDirs: () => youtubeFlowRuntime.cleanupSubtitleTempDirs(),
     cleanupYoutubeMediaCache: () => youtubeMediaCache.cleanup(),
+    cleanupRemoteMediaWindows: () => getSharedRemoteMediaWindowCache().cleanup(),
     cleanupJellyfinSubtitleCache: () => cleanupJellyfinSubtitleCache(),
     stopDiscordPresenceService: () => {
       void appState.discordPresenceService?.stop();
@@ -4572,7 +4582,6 @@ const {
           appState.activeParsedSubtitleSource = null;
           appState.activeParsedSubtitleMediaPath = null;
         }
-        activeJellyfinSubtitleDelayKey = null;
         overlayManager.broadcastToOverlayWindows('subtitle:set', resetSubtitlePayload);
         subtitleWsService.broadcast(resetSubtitlePayload, frequencyOptions);
         annotationSubtitleWsService.broadcast(resetSubtitlePayload, frequencyOptions);

src/main/overlay-runtime.test.ts

@@ -828,6 +828,7 @@ test('modal fallback reveal skips showing window when content is not ready', asy
       setModalWindowBounds: () => {},
     },
     {
+      platform: 'darwin',
       scheduleRevealFallback: (callback) => {
         scheduledReveal = callback;
         return { scheduled: true } as never;
@@ -1363,3 +1364,62 @@ test('modal placement reconcile cancels stale retry ladder after a newer visible
     globalThis.clearTimeout = originalClearTimeout;
   }
 });
+
+test('Linux keeps the dedicated modal window unmapped until the renderer opens the modal, then hides the overlay before revealing it', () => {
+  const mainWindow = createMockWindow();
+  mainWindow.visible = true;
+  const modalWindow = createMockWindow();
+  const order: string[] = [];
+  const hideMain = mainWindow.hide;
+  mainWindow.hide = () => {
+    order.push('main:hide');
+    hideMain();
+  };
+  const showModal = modalWindow.show;
+  modalWindow.show = () => {
+    order.push('modal:show');
+    showModal();
+  };
+  let revealScheduled = false;
+  const runtime = createOverlayModalRuntimeService(
+    {
+      getMainWindow: () => mainWindow as never,
+      getModalWindow: () => modalWindow as never,
+      createModalWindow: () => modalWindow as never,
+      getModalGeometry: () => ({ x: 0, y: 0, width: 400, height: 300 }),
+      setModalWindowBounds: () => {},
+    },
+    {
+      platform: 'linux',
+      scheduleRevealFallback: () => {
+        revealScheduled = true;
+        return { scheduled: true } as never;
+      },
+      clearRevealFallback: () => {},
+    },
+  );
+
+  const open = () =>
+    runtime.sendToActiveOverlayWindow(
+      'media-timing-review:open',
+      { reviewId: 'review' },
+      { restoreOnModalClose: 'media-timing-review', preferModalWindow: true },
+    );
+
+  assert.equal(open(), true);
+  assert.deepEqual(modalWindow.sent, [['media-timing-review:open', { reviewId: 'review' }]]);
+  assert.equal(revealScheduled, false);
+  assert.equal(modalWindow.getShowCount(), 0);
+  assert.equal(mainWindow.getHideCount(), 0);
+
+  // The open retry must not map the window before the renderer answers either.
+  assert.equal(open(), true);
+  assert.equal(modalWindow.getShowCount(), 0);
+
+  runtime.notifyOverlayModalOpened('media-timing-review');
+
+  assert.deepEqual(order, ['main:hide', 'modal:show']);
+  assert.equal(mainWindow.isVisible(), false);
+  assert.equal(modalWindow.isVisible(), true);
+  assert.equal(modalWindow.ignoreMouseEvents, false);
+});

src/main/overlay-runtime.ts

@@ -90,6 +90,12 @@ export function createOverlayModalRuntimeService(
   const platform = options.platform ?? process.platform;
   const shouldPrimeModalWindow = platform === 'darwin' || platform === 'win32';
   const reuseModalWindowAfterClose = platform === 'darwin';
+  // On Linux (Hyprland) every placement dispatch on a mapped window (resize, move, set_prop)
+  // blanks the still-visible overlay for a few frames while mpv is fullscreen. Revealing the
+  // dedicated modal window before its renderer has the modal open runs the placement ladder,
+  // and the open retry, against a visible overlay, which the user sees as flicker. Keep the
+  // window unmapped until the renderer acknowledges the open, then hide the overlay first.
+  const deferModalRevealUntilOpened = platform === 'linux';
   const focusApplication = options.focusApplication ?? requestOverlayApplicationFocus;
   const scheduleRevealFallback = (callback: () => void, delayMs: number): RevealFallbackHandle =>
     (options.scheduleRevealFallback ?? globalThis.setTimeout)(callback, delayMs);
@@ -457,7 +463,9 @@ export function createOverlayModalRuntimeService(
       deps.setModalWindowBounds(deps.getModalGeometry());
       const wasVisible = modalWindow.isVisible();
       if (!wasVisible) {
-        if (modalWindowPrimedForImmediateShow && isWindowReadyForIpc(modalWindow)) {
+        if (deferModalRevealUntilOpened) {
+          // notifyOverlayModalOpened reveals the window once the renderer has the modal open.
+        } else if (modalWindowPrimedForImmediateShow && isWindowReadyForIpc(modalWindow)) {
           showModalWindow(modalWindow);
         } else {
           scheduleModalWindowReveal(modalWindow);
@@ -560,15 +568,23 @@ export function createOverlayModalRuntimeService(
     }
 
     const modalWindow = deps.getModalWindow();
+    const targetIsModalWindow =
+      modalWindow !== null && !modalWindow.isDestroyed() && targetWindow === modalWindow;
+    const handOffMainWindowToModal = (): void => {
+      setMainWindowMousePassthroughForModal(true);
+      setMainWindowVisibilityForModal(true);
+    };
+
+    if (targetIsModalWindow && deferModalRevealUntilOpened) {
+      handOffMainWindowToModal();
+    }
     if (targetWindow.isVisible()) {
       ensureModalWindowInteractive(targetWindow);
     } else {
       showModalWindow(targetWindow);
     }
-
-    if (modalWindow && !modalWindow.isDestroyed() && targetWindow === modalWindow) {
-      setMainWindowMousePassthroughForModal(true);
-      setMainWindowVisibilityForModal(true);
+    if (targetIsModalWindow && !deferModalRevealUntilOpened) {
+      handOffMainWindowToModal();
     }
   };
 

src/main/runtime/app-lifecycle-actions.test.ts

@@ -46,12 +46,13 @@ test('on will quit cleanup handler runs all cleanup steps', () => {
     cleanupInternalSubtitleTrackCache: () => calls.push('cleanup-internal-subtitles'),
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
     cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
+    cleanupRemoteMediaWindows: () => calls.push('cleanup-remote-media-windows'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });
 
   cleanup();
-  assert.equal(calls.length, 35);
+  assert.equal(calls.length, 36);
   assert.equal(calls[0], 'destroy-tray');
   assert.equal(calls[calls.length - 1], 'stop-discord-presence');
   assert.ok(calls.includes('cleanup-jellyfin-subtitles'));
@@ -60,6 +61,7 @@ test('on will quit cleanup handler runs all cleanup steps', () => {
   assert.ok(calls.includes('clear-linux-mpv-fullscreen-overlay-refresh-timeouts'));
   assert.ok(calls.includes('cleanup-youtube-subtitles'));
   assert.ok(calls.includes('cleanup-youtube-media'));
+  assert.ok(calls.includes('cleanup-remote-media-windows'));
   assert.ok(calls.indexOf('flush-mpv-log') < calls.indexOf('destroy-socket'));
 });
 
@@ -102,6 +104,7 @@ test('on will quit cleanup handler cleans jellyfin subtitle cache when stopping
     cleanupInternalSubtitleTrackCache: () => calls.push('cleanup-internal-subtitles'),
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
     cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
+    cleanupRemoteMediaWindows: () => calls.push('cleanup-remote-media-windows'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });

src/main/runtime/app-lifecycle-actions.ts

@@ -32,6 +32,7 @@ export function createOnWillQuitCleanupHandler(deps: {
   cleanupInternalSubtitleTrackCache: () => void;
   cleanupYoutubeSubtitleTempDirs: () => void;
   cleanupYoutubeMediaCache: () => void;
+  cleanupRemoteMediaWindows: () => void;
   cleanupJellyfinSubtitleCache: () => void;
   stopDiscordPresenceService: () => void;
 }) {
@@ -76,6 +77,7 @@ export function createOnWillQuitCleanupHandler(deps: {
     }
     deps.cleanupYoutubeSubtitleTempDirs();
     deps.cleanupYoutubeMediaCache();
+    deps.cleanupRemoteMediaWindows();
     deps.stopDiscordPresenceService();
     return Promise.resolve(stopSyncAutoScheduler);
   };

src/main/runtime/app-lifecycle-main-cleanup.test.ts

@@ -75,6 +75,7 @@ test('cleanup deps builder returns handlers that guard optional runtime objects'
     cleanupInternalSubtitleTrackCache: () => calls.push('cleanup-internal-subtitles'),
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
     cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
+    cleanupRemoteMediaWindows: () => calls.push('cleanup-remote-media-windows'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });
@@ -157,6 +158,7 @@ test('cleanup deps builder skips destroyed yomitan window', () => {
     cleanupInternalSubtitleTrackCache: () => {},
     cleanupYoutubeSubtitleTempDirs: () => {},
     cleanupYoutubeMediaCache: () => {},
+    cleanupRemoteMediaWindows: () => {},
     cleanupJellyfinSubtitleCache: () => {},
     stopDiscordPresenceService: () => {},
   });
@@ -210,6 +212,7 @@ test('cleanup deps builder skips global shortcut cleanup before app ready', () =
     cleanupInternalSubtitleTrackCache: () => {},
     cleanupYoutubeSubtitleTempDirs: () => {},
     cleanupYoutubeMediaCache: () => {},
+    cleanupRemoteMediaWindows: () => {},
     cleanupJellyfinSubtitleCache: () => {},
     stopDiscordPresenceService: () => {},
   });

src/main/runtime/app-lifecycle-main-cleanup.ts

@@ -61,6 +61,7 @@ export function createBuildOnWillQuitCleanupDepsHandler(deps: {
   cleanupInternalSubtitleTrackCache: () => void;
   cleanupYoutubeSubtitleTempDirs: () => void;
   cleanupYoutubeMediaCache: () => void;
+  cleanupRemoteMediaWindows: () => void;
   cleanupJellyfinSubtitleCache: () => void;
   stopDiscordPresenceService: () => void;
 }) {
@@ -148,6 +149,7 @@ export function createBuildOnWillQuitCleanupDepsHandler(deps: {
     cleanupInternalSubtitleTrackCache: () => deps.cleanupInternalSubtitleTrackCache(),
     cleanupYoutubeSubtitleTempDirs: () => deps.cleanupYoutubeSubtitleTempDirs(),
     cleanupYoutubeMediaCache: () => deps.cleanupYoutubeMediaCache(),
+    cleanupRemoteMediaWindows: () => deps.cleanupRemoteMediaWindows(),
     cleanupJellyfinSubtitleCache: () => deps.cleanupJellyfinSubtitleCache(),
     stopDiscordPresenceService: () => deps.stopDiscordPresenceService(),
   });

src/main/runtime/composers/startup-lifecycle-composer.test.ts

@@ -52,6 +52,7 @@ test('composeStartupLifecycleHandlers returns callable startup lifecycle handler
       cleanupInternalSubtitleTrackCache: () => {},
       cleanupYoutubeSubtitleTempDirs: () => {},
       cleanupYoutubeMediaCache: () => {},
+      cleanupRemoteMediaWindows: () => {},
       cleanupJellyfinSubtitleCache: () => {},
       stopDiscordPresenceService: () => {},
     },

src/main/runtime/config-hot-reload-handlers.ts

@@ -137,6 +137,9 @@ function buildAnkiRuntimeConfigPatch(
   if (diff.hotReloadFields.includes('ankiConnect.isKiku.fieldGrouping')) {
     patch.isKiku = { fieldGrouping: config.ankiConnect.isKiku.fieldGrouping };
   }
+  if (diff.hotReloadFields.includes('ankiConnect.isSenren.fieldGrouping')) {
+    patch.isSenren = { fieldGrouping: config.ankiConnect.isSenren.fieldGrouping };
+  }
   if (diff.hotReloadFields.includes('ankiConnect.lapisKiku.wordCardKind')) {
     patch.lapisKiku = { wordCardKind: config.ankiConnect.lapisKiku.wordCardKind };
   }

src/main/runtime/jellyfin-subtitle-preload-main-deps.test.ts

@@ -19,19 +19,6 @@ test('preload jellyfin external subtitles main deps builder maps callbacks', asy
       return { path: '/tmp/sub.srt', cleanupDir: '/tmp/subs' };
     },
     cleanupCachedSubtitles: () => calls.push('cleanup'),
-    getSavedSubtitleDelay: (_itemId, streamIndex) => {
-      calls.push(`load-delay:${streamIndex}`);
-      return 1.25;
-    },
-    setActiveSubtitleDelayKey: (key) => calls.push(`active-delay:${key?.streamIndex ?? 'none'}`),
-    loadSubtitleSourceText: async (source) => {
-      calls.push(`load-source:${source}`);
-      return 'subtitle';
-    },
-    saveSubtitleDelay: (_itemId, streamIndex, delaySeconds) => {
-      calls.push(`save-delay:${streamIndex}:${delaySeconds}`);
-      return true;
-    },
     logDebug: (message) => calls.push(`debug:${message}`),
   })();
 
@@ -41,21 +28,6 @@ test('preload jellyfin external subtitles main deps builder maps callbacks', asy
   await deps.wait(1);
   await deps.cacheSubtitleTrack({ index: 1, deliveryUrl: 'https://example.test/sub.srt' });
   deps.cleanupCachedSubtitles(['/tmp/subs']);
-  assert.equal(deps.getSavedSubtitleDelay?.('item', 3), 1.25);
-  deps.setActiveSubtitleDelayKey?.({ itemId: 'item', streamIndex: 3 });
-  assert.equal(await deps.loadSubtitleSourceText?.('/tmp/sub.srt'), 'subtitle');
-  assert.equal(deps.saveSubtitleDelay?.('item', 3, -31.5), true);
   deps.logDebug('oops', null);
-  assert.deepEqual(calls, [
-    'list',
-    'send',
-    'wait',
-    'cache',
-    'cleanup',
-    'load-delay:3',
-    'active-delay:3',
-    'load-source:/tmp/sub.srt',
-    'save-delay:3:-31.5',
-    'debug:oops',
-  ]);
+  assert.deepEqual(calls, ['list', 'send', 'wait', 'cache', 'cleanup', 'debug:oops']);
 });

src/main/runtime/jellyfin-subtitle-preload-main-deps.ts

@@ -15,19 +15,6 @@ export function createBuildPreloadJellyfinExternalSubtitlesMainDepsHandler(
     wait: (ms: number) => deps.wait(ms),
     cacheSubtitleTrack: (track) => deps.cacheSubtitleTrack(track),
     cleanupCachedSubtitles: (dirs) => deps.cleanupCachedSubtitles(dirs),
-    getSavedSubtitleDelay: deps.getSavedSubtitleDelay
-      ? (itemId, streamIndex) => deps.getSavedSubtitleDelay!(itemId, streamIndex)
-      : undefined,
-    setActiveSubtitleDelayKey: deps.setActiveSubtitleDelayKey
-      ? (key) => deps.setActiveSubtitleDelayKey!(key)
-      : undefined,
-    loadSubtitleSourceText: deps.loadSubtitleSourceText
-      ? (source) => deps.loadSubtitleSourceText!(source)
-      : undefined,
-    saveSubtitleDelay: deps.saveSubtitleDelay
-      ? (itemId, streamIndex, delaySeconds) =>
-          deps.saveSubtitleDelay!(itemId, streamIndex, delaySeconds)
-      : undefined,
     initSubtitlePrefetch: deps.initSubtitlePrefetch
       ? (sourcePath) => deps.initSubtitlePrefetch!(sourcePath)
       : undefined,

src/main/runtime/jellyfin-subtitle-preload.test.ts

@@ -32,14 +32,6 @@ function makeDeps(overrides: {
   cleanupCachedSubtitles?: Parameters<
     typeof createPreloadJellyfinExternalSubtitlesHandler
   >[0]['cleanupCachedSubtitles'];
-  getSavedSubtitleDelay?: Parameters<
-    typeof createPreloadJellyfinExternalSubtitlesHandler
-  >[0]['getSavedSubtitleDelay'];
-  setActiveSubtitleDelayKey?: Parameters<
-    typeof createPreloadJellyfinExternalSubtitlesHandler
-  >[0]['setActiveSubtitleDelayKey'];
-  loadSubtitleSourceText?: (source: string) => Promise<string>;
-  saveSubtitleDelay?: (itemId: string, streamIndex: number, delaySeconds: number) => void;
   initSubtitlePrefetch?: Parameters<
     typeof createPreloadJellyfinExternalSubtitlesHandler
   >[0]['initSubtitlePrefetch'];
@@ -57,10 +49,6 @@ function makeDeps(overrides: {
         cleanupDir: '/tmp/subminer-jellyfin-subtitles',
       })),
     cleanupCachedSubtitles: overrides.cleanupCachedSubtitles ?? (() => {}),
-    getSavedSubtitleDelay: overrides.getSavedSubtitleDelay,
-    setActiveSubtitleDelayKey: overrides.setActiveSubtitleDelayKey,
-    loadSubtitleSourceText: overrides.loadSubtitleSourceText,
-    saveSubtitleDelay: overrides.saveSubtitleDelay,
     initSubtitlePrefetch: overrides.initSubtitlePrefetch,
     logDebug: overrides.logDebug ?? (() => {}),
   };
@@ -377,20 +365,17 @@ test('preload jellyfin subtitles waits for delayed external japanese track inste
 
 test('preload jellyfin subtitles clears managed delay when no external tracks are available', async () => {
   const commands: Array<Array<string | number>> = [];
-  const activeDelayKeys: Array<unknown> = [];
   const preload = createPreloadJellyfinExternalSubtitlesHandler(
     makeDeps({
       listJellyfinSubtitleTracks: async () => [
         { index: 0, language: 'jpn', title: 'Embedded Japanese' },
       ],
       sendMpvCommand: (command) => commands.push(command),
-      setActiveSubtitleDelayKey: (key) => activeDelayKeys.push(key),
     }),
   );
 
   await preload({ session, clientInfo, itemId: 'item-1' });
 
-  assert.deepEqual(activeDelayKeys, [null]);
   assert.deepEqual(commands, [['set_property', 'sub-delay', 0]]);
 });
 
@@ -461,42 +446,7 @@ test('preload jellyfin subtitles prefers Jellyfin default and embedded japanese
   ]);
 });
 
-test('preload jellyfin subtitles applies saved delay for selected japanese stream', async () => {
-  const commands: Array<Array<string | number>> = [];
-  const activeKeys: Array<{ itemId: string; streamIndex: number } | null> = [];
-  const preload = createPreloadJellyfinExternalSubtitlesHandler(
-    makeDeps({
-      listJellyfinSubtitleTracks: async () => [
-        { index: 3, language: 'jpn', title: 'Japanese', deliveryUrl: 'https://sub/jpn.srt' },
-      ],
-      getMpvClient: () => ({
-        requestProperty: async () => [
-          {
-            type: 'sub',
-            id: 11,
-            lang: 'jpn',
-            title: 'Japanese',
-            external: true,
-            'external-filename': '/tmp/subminer-jellyfin-subtitles/3.srt',
-          },
-        ],
-      }),
-      sendMpvCommand: (command) => commands.push(command),
-      getSavedSubtitleDelay: (_itemId, streamIndex) => (streamIndex === 3 ? 1.25 : null),
-      setActiveSubtitleDelayKey: (key) => activeKeys.push(key),
-    }),
-  );
-
-  await preload({ session, clientInfo, itemId: 'item-9' });
-
-  assert.deepEqual(setPropertyCommandsExceptTrackAutoSelection(commands), [
-    ['set_property', 'sub-delay', 1.25],
-    ['set_property', 'sid', 11],
-  ]);
-  assert.deepEqual(activeKeys, [{ itemId: 'item-9', streamIndex: 3 }]);
-});
-
-test('preload jellyfin subtitles applies saved delay before selecting japanese stream', async () => {
+test('preload jellyfin subtitles resets delay before selecting japanese stream', async () => {
   const commands: Array<Array<string | number>> = [];
   const preload = createPreloadJellyfinExternalSubtitlesHandler(
     makeDeps({
@@ -516,14 +466,13 @@ test('preload jellyfin subtitles applies saved delay before selecting japanese s
         ],
       }),
       sendMpvCommand: (command) => commands.push(command),
-      getSavedSubtitleDelay: () => 1.25,
     }),
   );
 
   await preload({ session, clientInfo, itemId: 'item-9' });
 
   const delayIndex = commands.findIndex(
-    (command) => command[0] === 'set_property' && command[1] === 'sub-delay' && command[2] === 1.25,
+    (command) => command[0] === 'set_property' && command[1] === 'sub-delay' && command[2] === 0,
   );
   const selectedSidIndex = commands.findIndex(
     (command) => command[0] === 'set_property' && command[1] === 'sid' && command[2] === 11,
@@ -533,143 +482,6 @@ test('preload jellyfin subtitles applies saved delay before selecting japanese s
   assert.ok(delayIndex < selectedSidIndex);
 });
 
-test('preload jellyfin subtitles auto-aligns late japanese track from english reference', async () => {
-  const commands: Array<Array<string | number>> = [];
-  const savedDelays: Array<{ itemId: string; streamIndex: number; delaySeconds: number }> = [];
-  const primarySrt = `1
-00:00:34,935 --> 00:00:36,937
-Japanese 1
-
-2
-00:00:36,937 --> 00:00:41,441
-Japanese 2
-
-3
-00:00:41,441 --> 00:00:45,279
-Japanese 3
-
-4
-00:00:45,279 --> 00:00:48,115
-Japanese 4
-
-5
-00:00:48,115 --> 00:00:52,286
-Japanese 5
-
-6
-00:00:52,286 --> 00:00:54,955
-Japanese 6
-
-7
-00:00:54,955 --> 00:00:59,793
-Japanese 7
-
-8
-00:00:59,793 --> 00:01:03,630
-Japanese 8
-
-9
-00:01:03,630 --> 00:01:07,634
-Japanese 9
-
-10
-00:01:07,634 --> 00:01:13,040
-Japanese 10
-
-11
-00:01:16,643 --> 00:01:20,814
-Japanese 11
-
-12
-00:01:20,814 --> 00:01:23,116
-Japanese 12
-
-13
-00:01:27,988 --> 00:01:30,991
-Japanese 13
-
-14
-00:01:30,991 --> 00:01:34,094
-Japanese 14
-
-15
-00:01:34,094 --> 00:01:37,097
-Japanese 15
-
-16
-00:01:37,097 --> 00:01:39,100
-Japanese 16
-`;
-  const referenceAss = `[Events]
-Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
-Dialogue: 0,0:00:03.46,0:00:08.73,Default,,0,0,0,,English 1
-Dialogue: 0,0:00:09.48,0:00:13.61,Default,,0,0,0,,English 2
-Dialogue: 0,0:00:13.61,0:00:19.64,Default,,0,0,0,,English 3
-Dialogue: 0,0:00:21.40,0:00:27.32,Default,,0,0,0,,English 4
-Dialogue: 0,0:00:28.16,0:00:31.75,Default,,0,0,0,,English 5
-Dialogue: 0,0:00:32.06,0:00:34.52,Default,,0,0,0,,English 6
-Dialogue: 0,0:00:35.93,0:00:40.57,Default,,0,0,0,,English 7
-Dialogue: 0,0:00:45.10,0:00:51.01,Default,,0,0,0,,English 8
-Dialogue: 0,0:00:56.57,0:00:59.12,Default,,0,0,0,,English 9
-Dialogue: 0,0:00:59.68,0:01:02.44,Default,,0,0,0,,English 10
-Dialogue: 0,0:01:02.44,0:01:05.56,Default,,0,0,0,,English 11
-Dialogue: 0,0:01:05.56,0:01:06.87,Default,,0,0,0,,English 12
-`;
-  const preload = createPreloadJellyfinExternalSubtitlesHandler(
-    makeDeps({
-      listJellyfinSubtitleTracks: async () => [
-        { index: 0, language: 'jpn', title: 'Japanese', deliveryUrl: 'https://sub/jpn.srt' },
-        { index: 4, language: 'eng', title: 'English', deliveryUrl: 'https://sub/eng.ass' },
-      ],
-      getMpvClient: () => ({
-        requestProperty: async () => [
-          {
-            type: 'sub',
-            id: 10,
-            lang: 'jpn',
-            title: 'Japanese',
-            external: true,
-            'external-filename': '/tmp/subminer-jellyfin-subtitles/0.srt',
-          },
-          {
-            type: 'sub',
-            id: 12,
-            lang: 'eng',
-            title: 'English',
-            external: true,
-            'external-filename': '/tmp/subminer-jellyfin-subtitles/4.ass',
-          },
-        ],
-      }),
-      sendMpvCommand: (command) => commands.push(command),
-      cacheSubtitleTrack: async (track) => ({
-        path: `/tmp/subminer-jellyfin-subtitles/${track.index}.${track.index === 4 ? 'ass' : 'srt'}`,
-        cleanupDir: '/tmp/subminer-jellyfin-subtitles',
-      }),
-      getSavedSubtitleDelay: () => null,
-      loadSubtitleSourceText: async (source) =>
-        source.endsWith('.ass') ? referenceAss : primarySrt,
-      saveSubtitleDelay: (itemId, streamIndex, delaySeconds) => {
-        savedDelays.push({ itemId, streamIndex, delaySeconds });
-      },
-    }),
-  );
-
-  await preload({ session, clientInfo, itemId: 'item-9' });
-
-  const delayCommand = commands.find(
-    (command) => command[0] === 'set_property' && command[1] === 'sub-delay',
-  );
-  assert.ok(delayCommand);
-  const delaySeconds = delayCommand[2];
-  if (typeof delaySeconds !== 'number') {
-    assert.fail('Expected numeric subtitle delay.');
-  }
-  assert.ok(delaySeconds > -32);
-  assert.ok(delaySeconds < -31);
-  assert.deepEqual(savedDelays, [{ itemId: 'item-9', streamIndex: 0, delaySeconds }]);
-});
-
 test('preload jellyfin subtitles accepts numeric string mpv track ids', async () => {
   const commands: Array<Array<string | number>> = [];
   const preload = createPreloadJellyfinExternalSubtitlesHandler(

src/main/runtime/jellyfin-subtitle-preload.ts

@@ -1,6 +1,3 @@
-import { parseSubtitleCues } from '../../core/services/subtitle-cue-parser';
-import { estimateSubtitleTimingOffset } from '../../core/services/subtitle-timing-offset';
-
 type JellyfinSession = {
   serverUrl: string;
   accessToken: string;
@@ -35,11 +32,6 @@ type CachedExternalSubtitleTrack = CachedSubtitleTrack & {
   source: JellyfinSubtitleTrack;
 };
 
-type JellyfinSubtitleDelayKey = {
-  itemId: string;
-  streamIndex: number;
-};
-
 type MpvSubtitleTrack = {
   id: number;
   lang: string;
@@ -257,54 +249,6 @@ async function waitForPreferredSubtitleTracks(
   return subtitleTracks;
 }
 
-async function estimateSubtitleDelayFromReference(
-  deps: {
-    loadSubtitleSourceText?: (source: string) => Promise<string>;
-    logDebug: (message: string, error: unknown) => void;
-  },
-  primaryTrack: CachedExternalSubtitleTrack | null,
-  referenceTrack: CachedExternalSubtitleTrack | null,
-): Promise<number | null> {
-  if (!deps.loadSubtitleSourceText || !primaryTrack || !referenceTrack) {
-    return null;
-  }
-
-  try {
-    const [primaryContent, referenceContent] = await Promise.all([
-      deps.loadSubtitleSourceText(primaryTrack.path),
-      deps.loadSubtitleSourceText(referenceTrack.path),
-    ]);
-    const primaryCues = parseSubtitleCues(primaryContent, primaryTrack.path);
-    const referenceCues = parseSubtitleCues(referenceContent, referenceTrack.path);
-    return estimateSubtitleTimingOffset(primaryCues, referenceCues)?.offsetSeconds ?? null;
-  } catch (error) {
-    deps.logDebug('Failed to auto-align Jellyfin subtitle timing', error);
-    return null;
-  }
-}
-
-function saveEstimatedSubtitleDelay(
-  deps: {
-    saveSubtitleDelay?: (
-      itemId: string,
-      streamIndex: number,
-      delaySeconds: number,
-    ) => boolean | void;
-    logDebug: (message: string, error: unknown) => void;
-  },
-  key: JellyfinSubtitleDelayKey,
-  delaySeconds: number,
-): void {
-  try {
-    const saved = deps.saveSubtitleDelay?.(key.itemId, key.streamIndex, delaySeconds);
-    if (saved === false) {
-      deps.logDebug('Failed to save Jellyfin auto subtitle delay', key);
-    }
-  } catch (error) {
-    deps.logDebug('Failed to save Jellyfin auto subtitle delay', error);
-  }
-}
-
 export function createPreloadJellyfinExternalSubtitlesHandler(deps: {
   listJellyfinSubtitleTracks: (
     session: JellyfinSession,
@@ -316,10 +260,6 @@ export function createPreloadJellyfinExternalSubtitlesHandler(deps: {
   wait: (ms: number) => Promise<void>;
   cacheSubtitleTrack: (track: JellyfinSubtitleTrack) => Promise<CachedSubtitleTrack>;
   cleanupCachedSubtitles: (dirs: string[]) => void;
-  getSavedSubtitleDelay?: (itemId: string, streamIndex: number) => number | null;
-  setActiveSubtitleDelayKey?: (key: JellyfinSubtitleDelayKey | null) => void;
-  loadSubtitleSourceText?: (source: string) => Promise<string>;
-  saveSubtitleDelay?: (itemId: string, streamIndex: number, delaySeconds: number) => boolean | void;
   initSubtitlePrefetch?: (sourcePath: string) => void | Promise<void>;
   logDebug: (message: string, error: unknown) => void;
 }): PreloadJellyfinExternalSubtitlesHandler {
@@ -357,6 +297,7 @@ export function createPreloadJellyfinExternalSubtitlesHandler(deps: {
     itemId: string;
   }): Promise<void> => {
     try {
+      resetManagedSubtitleDelay();
       try {
         cleanupActiveCache();
       } catch (error) {
@@ -369,8 +310,6 @@ export function createPreloadJellyfinExternalSubtitlesHandler(deps: {
       );
       const externalTracks = tracks.filter((track) => Boolean(track.deliveryUrl));
       if (externalTracks.length === 0) {
-        deps.setActiveSubtitleDelayKey?.(null);
-        resetManagedSubtitleDelay();
         return;
       }
 
@@ -427,40 +366,13 @@ export function createPreloadJellyfinExternalSubtitlesHandler(deps: {
           japanesePrimaryId,
         );
         if (selectedCachedTrack) {
-          const delayKey = { itemId: params.itemId, streamIndex: selectedCachedTrack.source.index };
-          deps.setActiveSubtitleDelayKey?.(delayKey);
-          const savedDelay = deps.getSavedSubtitleDelay?.(delayKey.itemId, delayKey.streamIndex);
-          if (typeof savedDelay === 'number' && Number.isFinite(savedDelay)) {
-            deps.sendMpvCommand(['set_property', 'sub-delay', savedDelay]);
-          } else {
-            const referenceCachedTrack = findCachedTrackForMpvTrackId(
-              resolvedSubtitleTracks,
-              cachedTracks,
-              englishSecondaryId,
-            );
-            const estimatedDelay = await estimateSubtitleDelayFromReference(
-              deps,
-              selectedCachedTrack,
-              referenceCachedTrack,
-            );
-            if (estimatedDelay !== null) {
-              deps.sendMpvCommand(['set_property', 'sub-delay', estimatedDelay]);
-              saveEstimatedSubtitleDelay(deps, delayKey, estimatedDelay);
-            } else {
-              resetManagedSubtitleDelay();
-            }
-          }
           deps.sendMpvCommand(['set_property', 'sid', japanesePrimaryId]);
           startSubtitlePrefetchForCachedTrack(selectedCachedTrack.path);
         } else {
-          deps.setActiveSubtitleDelayKey?.(null);
-          resetManagedSubtitleDelay();
           deps.sendMpvCommand(['set_property', 'sid', japanesePrimaryId]);
         }
       } else {
         deps.sendMpvCommand(['set_property', 'sid', 'no']);
-        deps.setActiveSubtitleDelayKey?.(null);
-        resetManagedSubtitleDelay();
       }
 
       if (englishSecondaryId !== null) {

src/main/runtime/media-timing-review-open.ts

@@ -25,7 +25,9 @@ export async function openMediaTimingReviewModal(
     { waitForModalOpen: deps.waitForModalOpen, logWarn: deps.logWarn },
     {
       modal: MODAL,
-      timeoutMs: 1_500,
+      // The review renderer regularly needs more than the 1.5 s the other modals allow; a
+      // premature retry re-sends the payload and reloads the waveform for nothing.
+      timeoutMs: 4_000,
       retryWarning:
         'Media timing review did not acknowledge modal open; retrying the dedicated modal window.',
       sendOpen: () =>

src/main/runtime/media-timing-review.test.ts

@@ -1,6 +1,15 @@
 import assert from 'node:assert/strict';
 import { describe, test } from 'node:test';
 import type { MediaTimingReviewOpenPayload } from '../../types/anki';
+import type { SpeechWaveformOptions } from '../../core/services/media-timing-waveform';
+import type {
+  RemoteMediaWindow,
+  RemoteMediaWindowRange,
+  RemoteMediaWindowSource,
+} from '../../core/services/remote-media-window-cache';
+import type { MediaTimingPreviewSession } from '../../core/services/media-timing-preview';
+
+type MediaTimingPreviewSessionLike = Pick<MediaTimingPreviewSession, 'start'>;
 import {
   buildMediaTimingReviewPayload,
   collectMediaTimingContextLines,
@@ -69,6 +78,8 @@ async function startActiveMediaTimingReview(
   options: {
     maxMediaDuration?: number;
     decisionTimeoutMs?: number;
+    generateWaveform?: () => Promise<number[]>;
+    play?: () => Promise<void>;
   } = {},
 ) {
   const previewCalls: Array<[number, number]> = [];
@@ -85,14 +96,16 @@ async function startActiveMediaTimingReview(
     }),
     getCurrentMediaPath: () => '/video/show.mkv',
     getMpvExecutablePath: () => 'mpv',
-    generateWaveform: async () => [],
+    generateWaveform: options.generateWaveform ?? (async () => []),
     decisionTimeoutMs: options.decisionTimeoutMs,
     createPreviewSession: () => ({
       start: async () => undefined,
       play: async (startTime, endTime) => {
         previewCalls.push([startTime, endTime]);
+        await options.play?.();
       },
       stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
       dispose: () => undefined,
     }),
     openModal: async (payload) => {
@@ -136,6 +149,7 @@ test('media timing review pauses playback, resolves exact timing, and restores p
         previewCalls.push([startTime, endTime]);
       },
       stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
       dispose: () => undefined,
     }),
     openModal: async (payload) => {
@@ -177,12 +191,7 @@ test('media timing review pauses playback, resolves exact timing, and restores p
 });
 
 test('media timing review analyzes the visible range on the selected audio stream', async () => {
-  const waveformCalls: Array<{
-    mediaPath: string;
-    startTime: number;
-    endTime: number;
-    audioStreamIndex?: number;
-  }> = [];
+  const waveformCalls: SpeechWaveformOptions[] = [];
   let runtime: ReturnType<typeof createMediaTimingReviewRuntime>;
   runtime = createMediaTimingReviewRuntime({
     getMpvClient: () => ({
@@ -202,6 +211,7 @@ test('media timing review analyzes the visible range on the selected audio strea
       start: async () => undefined,
       play: async () => undefined,
       stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
       dispose: () => undefined,
     }),
     openModal: async (payload) => {
@@ -239,6 +249,302 @@ test('media timing review analyzes the visible range on the selected audio strea
   ]);
 });
 
+const REMOTE_STREAM_URL = 'https://jellyfin.example/Videos/abc/stream?static=true';
+
+function createWindowStub(options: { fail?: boolean } = {}) {
+  const calls: Array<{ source: RemoteMediaWindowSource; range: RemoteMediaWindowRange }> = [];
+  const acquireMediaWindow = async (
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ): Promise<RemoteMediaWindow> => {
+    calls.push({ source, range });
+    if (options.fail) throw new Error('offline');
+    const windowPath = `/tmp/window-${range.startTime}-${range.endTime}.mkv`;
+    return {
+      path: windowPath,
+      startTime: range.startTime,
+      endTime: range.endTime,
+      sourcePath: source.path,
+      audioStreamIndex: source.audioStreamIndex ?? null,
+      media: {
+        path: windowPath,
+        source: 'remote-window',
+        singleResolvedStream: true,
+        absoluteTimestamps: true,
+      },
+    };
+  };
+  return { calls, acquireMediaWindow };
+}
+
+function createRemoteReviewRuntime(options: {
+  windowStub: ReturnType<typeof createWindowStub>;
+  waveformCalls: SpeechWaveformOptions[];
+  previewStarts: Array<Parameters<MediaTimingPreviewSessionLike['start']>[0]>;
+  previewPlays: Array<[string, number, number]>;
+  disposed: string[];
+  openModal: (
+    runtime: ReturnType<typeof createMediaTimingReviewRuntime>,
+    payload: MediaTimingReviewOpenPayload,
+  ) => Promise<void>;
+}) {
+  let runtime!: ReturnType<typeof createMediaTimingReviewRuntime>;
+  runtime = createMediaTimingReviewRuntime({
+    getMpvClient: () => ({
+      connected: true,
+      currentVideoPath: REMOTE_STREAM_URL,
+      currentAudioStreamIndex: 2,
+      requestProperty: async (name) =>
+        ({ pause: true, duration: 100, aid: 3, volume: 60 })[
+          name as 'pause' | 'duration' | 'aid' | 'volume'
+        ] ?? null,
+      send: () => undefined,
+    }),
+    getCurrentMediaPath: () => REMOTE_STREAM_URL,
+    getMpvExecutablePath: () => 'mpv',
+    resolveMediaSource: async () => ({
+      path: REMOTE_STREAM_URL,
+      inputOptions: { reconnect: true },
+    }),
+    acquireMediaWindow: options.windowStub.acquireMediaWindow,
+    generateWaveform: async (waveformOptions) => {
+      options.waveformCalls.push(waveformOptions);
+      return [0.1, 0.8, 0.2];
+    },
+    createPreviewSession: () => {
+      let mediaPath = '';
+      return {
+        start: async (startOptions) => {
+          mediaPath = startOptions.mediaPath;
+          options.previewStarts.push(startOptions);
+        },
+        play: async (startTime, endTime) => {
+          options.previewPlays.push([mediaPath, startTime, endTime]);
+        },
+        stop: async () => undefined,
+        onPlaybackEnded: () => undefined,
+        dispose: () => {
+          options.disposed.push(mediaPath);
+        },
+      };
+    },
+    openModal: async (payload) => {
+      await options.openModal(runtime, payload);
+      return true;
+    },
+    showStatus: () => undefined,
+  });
+  return runtime;
+}
+
+test('media timing review downloads one window of a remote stream for the waveform and preview', async () => {
+  const windowStub = createWindowStub();
+  const waveformCalls: SpeechWaveformOptions[] = [];
+  const previewStarts: Array<Parameters<MediaTimingPreviewSessionLike['start']>[0]> = [];
+  const previewPlays: Array<[string, number, number]> = [];
+  const disposed: string[] = [];
+  const runtime = createRemoteReviewRuntime({
+    windowStub,
+    waveformCalls,
+    previewStarts,
+    previewPlays,
+    disposed,
+    openModal: async (active, payload) => {
+      const waveform = await active.getWaveform({
+        reviewId: payload.reviewId,
+        startTime: payload.timelineStartTime,
+        endTime: payload.timelineEndTime,
+      });
+      assert.deepEqual(waveform, { ok: true, peaks: [0.1, 0.8, 0.2] });
+      assert.deepEqual(
+        await active.previewRange({ reviewId: payload.reviewId, startTime: 9.5, endTime: 12.5 }),
+        { ok: true },
+      );
+      active.resolveReview({
+        reviewId: payload.reviewId,
+        decision: { action: 'confirm', startTime: 9.5, endTime: 12.5 },
+      });
+    },
+  });
+
+  const decision = await runtime.requestReview({
+    kind: 'word',
+    text: '字幕',
+    startTime: 10,
+    endTime: 12,
+    audioPadding: 0.5,
+    maxMediaDuration: 30,
+  });
+
+  assert.deepEqual(decision, { action: 'confirm', startTime: 9.5, endTime: 12.5 });
+  assert.deepEqual(windowStub.calls, [
+    {
+      source: { path: REMOTE_STREAM_URL, inputOptions: { reconnect: true }, audioStreamIndex: 2 },
+      range: { startTime: 7.5, endTime: 14.5 },
+    },
+  ]);
+  assert.deepEqual(waveformCalls, [
+    {
+      mediaPath: {
+        path: '/tmp/window-7.5-14.5.mkv',
+        source: 'remote-window',
+        singleResolvedStream: true,
+        absoluteTimestamps: true,
+      },
+      startTime: 7.5,
+      endTime: 14.5,
+    },
+  ]);
+  assert.deepEqual(previewStarts, [
+    {
+      mediaPath: '/tmp/window-7.5-14.5.mkv',
+      executablePath: 'mpv',
+      volume: 60,
+      absoluteTimestamps: true,
+    },
+  ]);
+  assert.deepEqual(previewPlays, [['/tmp/window-7.5-14.5.mkv', 9.5, 12.5]]);
+  assert.deepEqual(disposed, ['/tmp/window-7.5-14.5.mkv']);
+});
+
+test('media timing review restarts the preview on a wider window when the timeline grows', async () => {
+  const windowStub = createWindowStub();
+  const waveformCalls: SpeechWaveformOptions[] = [];
+  const previewStarts: Array<Parameters<MediaTimingPreviewSessionLike['start']>[0]> = [];
+  const previewPlays: Array<[string, number, number]> = [];
+  const disposed: string[] = [];
+  const runtime = createRemoteReviewRuntime({
+    windowStub,
+    waveformCalls,
+    previewStarts,
+    previewPlays,
+    disposed,
+    openModal: async (active, payload) => {
+      await active.previewRange({ reviewId: payload.reviewId, startTime: 9.5, endTime: 12.5 });
+      // The user revealed two more seconds before the clip.
+      await active.getWaveform({ reviewId: payload.reviewId, startTime: 5.5, endTime: 14.5 });
+      await active.previewRange({ reviewId: payload.reviewId, startTime: 6, endTime: 12.5 });
+      active.resolveReview({ reviewId: payload.reviewId, decision: { action: 'use-original' } });
+    },
+  });
+
+  await runtime.requestReview({
+    kind: 'sentence',
+    text: '字幕',
+    startTime: 10,
+    endTime: 12,
+    audioPadding: 0.5,
+    maxMediaDuration: 30,
+  });
+
+  assert.deepEqual(
+    windowStub.calls.map((call) => call.range),
+    [
+      { startTime: 7.5, endTime: 14.5 },
+      { startTime: 5.5, endTime: 14.5 },
+    ],
+  );
+  assert.deepEqual(
+    previewStarts.map((start) => start.mediaPath),
+    ['/tmp/window-7.5-14.5.mkv', '/tmp/window-5.5-14.5.mkv'],
+  );
+  assert.deepEqual(previewPlays, [
+    ['/tmp/window-7.5-14.5.mkv', 9.5, 12.5],
+    ['/tmp/window-5.5-14.5.mkv', 6, 12.5],
+  ]);
+  assert.deepEqual(disposed, ['/tmp/window-7.5-14.5.mkv', '/tmp/window-5.5-14.5.mkv']);
+  assert.equal(waveformCalls[0]?.startTime, 5.5);
+});
+
+test('media timing review falls back to the remote stream after one failed window download', async () => {
+  const windowStub = createWindowStub({ fail: true });
+  const waveformCalls: SpeechWaveformOptions[] = [];
+  const previewStarts: Array<Parameters<MediaTimingPreviewSessionLike['start']>[0]> = [];
+  const previewPlays: Array<[string, number, number]> = [];
+  const disposed: string[] = [];
+  const runtime = createRemoteReviewRuntime({
+    windowStub,
+    waveformCalls,
+    previewStarts,
+    previewPlays,
+    disposed,
+    openModal: async (active, payload) => {
+      await active.getWaveform({
+        reviewId: payload.reviewId,
+        startTime: payload.timelineStartTime,
+        endTime: payload.timelineEndTime,
+      });
+      await active.previewRange({ reviewId: payload.reviewId, startTime: 9.5, endTime: 12.5 });
+      active.resolveReview({ reviewId: payload.reviewId, decision: { action: 'use-original' } });
+    },
+  });
+
+  await runtime.requestReview({
+    kind: 'word',
+    text: '字幕',
+    startTime: 10,
+    endTime: 12,
+    audioPadding: 0.5,
+    maxMediaDuration: 30,
+  });
+
+  assert.equal(windowStub.calls.length, 1);
+  assert.deepEqual(waveformCalls, [
+    {
+      mediaPath: { path: REMOTE_STREAM_URL, inputOptions: { reconnect: true } },
+      startTime: 7.5,
+      endTime: 14.5,
+      audioStreamIndex: 2,
+    },
+  ]);
+  assert.deepEqual(previewStarts, [
+    { mediaPath: REMOTE_STREAM_URL, executablePath: 'mpv', volume: 60, audioTrackId: 3 },
+  ]);
+  assert.deepEqual(previewPlays, [[REMOTE_STREAM_URL, 9.5, 12.5]]);
+});
+
+test('media timing review never downloads windows for local media', async () => {
+  const windowStub = createWindowStub();
+  let runtime!: ReturnType<typeof createMediaTimingReviewRuntime>;
+  runtime = createMediaTimingReviewRuntime({
+    getMpvClient: () => ({
+      connected: true,
+      currentVideoPath: '/video/show.mkv',
+      requestProperty: async (name) => (name === 'duration' ? 100 : name === 'pause' ? true : null),
+      send: () => undefined,
+    }),
+    getCurrentMediaPath: () => '/video/show.mkv',
+    getMpvExecutablePath: () => 'mpv',
+    resolveMediaSource: async () => ({ path: '/video/show.mkv' }),
+    acquireMediaWindow: windowStub.acquireMediaWindow,
+    generateWaveform: async () => [0.1, 0.8, 0.2],
+    createPreviewSession: () => ({
+      start: async () => undefined,
+      play: async () => undefined,
+      stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
+      dispose: () => undefined,
+    }),
+    openModal: async (payload) => {
+      await runtime.getWaveform({ reviewId: payload.reviewId, startTime: 7.5, endTime: 14.5 });
+      runtime.resolveReview({ reviewId: payload.reviewId, decision: { action: 'use-original' } });
+      return true;
+    },
+    showStatus: () => undefined,
+  });
+
+  await runtime.requestReview({
+    kind: 'sentence',
+    text: '字幕',
+    startTime: 10,
+    endTime: 12,
+    audioPadding: 0.5,
+    maxMediaDuration: 30,
+  });
+
+  assert.equal(windowStub.calls.length, 0);
+});
+
 test('media timing review rejects stale and out-of-range actions before allowing discard', async () => {
   const { runtime, payload, pendingDecision, previewCalls } = await startActiveMediaTimingReview({
     maxMediaDuration: 3,
@@ -246,14 +552,14 @@ test('media timing review rejects stale and out-of-range actions before allowing
 
   assert.deepEqual(
     await runtime.previewRange({ reviewId: 'stale-review', startTime: 10, endTime: 12 }),
-    { ok: false, message: 'This timing review is no longer active.' },
+    { ok: false, stale: true, message: 'This timing review is no longer active.' },
   );
   assert.deepEqual(
     runtime.resolveReview({
       reviewId: 'stale-review',
       decision: { action: 'confirm', startTime: 10, endTime: 12 },
     }),
-    { ok: false, message: 'This timing review is no longer active.' },
+    { ok: false, stale: true, message: 'This timing review is no longer active.' },
   );
   assert.deepEqual(
     runtime.resolveReview({
@@ -346,6 +652,7 @@ test('media timing review does not resume playback when the prior state is unava
       },
       play: async () => undefined,
       stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
       dispose: () => undefined,
     }),
     openModal: async (payload) => {
@@ -384,11 +691,17 @@ test('media timing review restores playback when setup fails after pausing', asy
       send: ({ command }) => commands.push(command),
     }),
     getCurrentMediaPath: () => '/video/show.mkv',
-    getMpvExecutablePath: () => 'mpv',
-    generateWaveform: async () => [],
-    createPreviewSession: () => {
+    getMpvExecutablePath: () => {
       throw new Error('preview setup failed');
     },
+    generateWaveform: async () => [],
+    createPreviewSession: () => ({
+      start: async () => undefined,
+      play: async () => undefined,
+      stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
+      dispose: () => undefined,
+    }),
     openModal: async () => true,
     showStatus: () => undefined,
   });
@@ -426,6 +739,7 @@ test('disposing an open review settles it with original timing and restores play
       start: async () => undefined,
       play: async () => undefined,
       stop: async () => undefined,
+      onPlaybackEnded: () => undefined,
       dispose: () => undefined,
     }),
     openModal: async () => true,
@@ -449,3 +763,97 @@ test('disposing an open review settles it with original timing and restores play
     ['set_property', 'pause', 'no'],
   ]);
 });
+
+test('media timing review forwards the hidden player finishing a preview to the modal', async () => {
+  const endedReviewIds: string[] = [];
+  const playback: { ended?: () => void } = {};
+  let publishPayload!: (payload: MediaTimingReviewOpenPayload) => void;
+  const openedPayload = new Promise<MediaTimingReviewOpenPayload>((resolve) => {
+    publishPayload = resolve;
+  });
+  const runtime = createMediaTimingReviewRuntime({
+    getMpvClient: () => ({
+      connected: true,
+      currentVideoPath: '/video/show.mkv',
+      requestProperty: async (name) => (name === 'duration' ? 100 : null),
+      send: () => undefined,
+    }),
+    getCurrentMediaPath: () => '/video/show.mkv',
+    getMpvExecutablePath: () => 'mpv',
+    generateWaveform: async () => [],
+    createPreviewSession: () => ({
+      start: async () => undefined,
+      play: async () => undefined,
+      stop: async () => undefined,
+      onPlaybackEnded: (listener) => {
+        playback.ended = listener;
+      },
+      dispose: () => undefined,
+    }),
+    openModal: async (payload) => {
+      publishPayload(payload);
+      return true;
+    },
+    onPreviewEnded: (reviewId) => {
+      endedReviewIds.push(reviewId);
+    },
+    showStatus: () => undefined,
+  });
+  const pendingDecision = runtime.requestReview({
+    kind: 'sentence',
+    text: '字幕',
+    startTime: 10,
+    endTime: 12,
+    audioPadding: 0,
+    maxMediaDuration: 30,
+  });
+  const payload = await openedPayload;
+
+  assert.deepEqual(
+    await runtime.previewRange({ reviewId: payload.reviewId, startTime: 10, endTime: 12 }),
+    {
+      ok: true,
+    },
+  );
+  assert.ok(playback.ended);
+  playback.ended();
+  assert.deepEqual(endedReviewIds, [payload.reviewId]);
+
+  runtime.resolveReview({ reviewId: payload.reviewId, decision: { action: 'use-original' } });
+  await pendingDecision;
+  playback.ended();
+  assert.deepEqual(endedReviewIds, [payload.reviewId]);
+});
+
+test('preview reports a stale review when the review ends during playback', async () => {
+  let endReview: (() => Promise<void>) | null = null;
+  const { runtime, payload, pendingDecision } = await startActiveMediaTimingReview({
+    play: async () => {
+      await endReview?.();
+    },
+  });
+  endReview = () => runtime.dispose();
+
+  assert.deepEqual(
+    await runtime.previewRange({ reviewId: payload.reviewId, startTime: 10, endTime: 12 }),
+    { ok: false, stale: true, message: 'This timing review is no longer active.' },
+  );
+  await pendingDecision;
+});
+
+test('waveform reports a stale review when the review ends during analysis', async () => {
+  let endReview: (() => Promise<void>) | null = null;
+  const { runtime, payload, pendingDecision } = await startActiveMediaTimingReview({
+    generateWaveform: async () => {
+      await endReview?.();
+      return [0.1, 0.9, 0.2];
+    },
+  });
+  endReview = () => runtime.dispose();
+
+  assert.deepEqual(
+    await runtime.getWaveform({ reviewId: payload.reviewId, startTime: 8, endTime: 14 }),
+    { ok: false, stale: true, message: 'This timing review is no longer active.' },
+  );
+  await pendingDecision;
+});

src/main/runtime/media-timing-review.ts

@@ -11,6 +11,13 @@ import type {
   MediaTimingReviewWaveformResult,
 } from '../../types/anki';
 import type { SpeechWaveformOptions } from '../../core/services/media-timing-waveform';
+import {
+  isRemoteMediaWindowSourcePath,
+  type RemoteMediaWindow,
+  type RemoteMediaWindowRange,
+  type RemoteMediaWindowSource,
+} from '../../core/services/remote-media-window-cache';
+import type { MediaInput, MediaInputOptions } from '../../media-input';
 
 const INITIAL_TIMELINE_MARGIN_SECONDS = 2;
 const REVIEW_DECISION_TIMEOUT_MS = 5 * 60_000;
@@ -31,19 +38,38 @@ interface PreviewSession {
     executablePath?: string;
     audioTrackId?: number;
     volume?: number;
+    absoluteTimestamps?: boolean;
   }): Promise<void>;
   play(startTime: number, endTime: number): Promise<void>;
   stop(): Promise<void>;
+  /** Fires when the player reaches the end of the clip started by play(). */
+  onPlaybackEnded(listener: () => void): void;
   dispose(): void;
 }
 
+interface ReviewMediaSource {
+  path: string;
+  inputOptions?: MediaInputOptions;
+  singleResolvedStream?: boolean;
+}
+
 interface ActiveReview {
   payload: MediaTimingReviewOpenPayload;
+  /** What the hidden mpv preview plays when no cached window is available. */
   mediaPath: string;
+  /** What the waveform reads when no cached window is available. */
+  waveformMedia: MediaInput;
   audioStreamIndex?: number;
+  /** Remote source to download windows of; null for local media or without a cache. */
+  windowSource: RemoteMediaWindowSource | null;
+  /** Latest window returned for this review; reused while it still covers the request. */
+  window: RemoteMediaWindow | null;
+  windowRequest: (RemoteMediaWindowRange & { promise: Promise<RemoteMediaWindow | null> }) | null;
+  windowFailed: boolean;
+  previewOptions: { executablePath?: string; audioTrackId?: number; volume?: number };
+  preview: { path: string; session: Promise<PreviewSession> } | null;
   mpvClient: ReviewMpvClient;
   restorePlayback: boolean;
-  preview: Promise<PreviewSession>;
   resolve: (decision: MediaTimingReviewDecision) => void;
 }
 
@@ -53,12 +79,21 @@ export interface MediaTimingReviewRuntimeDeps {
   getMpvExecutablePath: () => string;
   createPreviewSession: () => PreviewSession;
   generateWaveform: (options: SpeechWaveformOptions) => Promise<number[]>;
+  /** Resolves the FFmpeg-readable stream URL and headers behind the current media path. */
+  resolveMediaSource?: () => Promise<ReviewMediaSource | null>;
+  /** Downloads (or reuses) a local window of a remote source covering the range. */
+  acquireMediaWindow?: (
+    source: RemoteMediaWindowSource,
+    range: RemoteMediaWindowRange,
+  ) => Promise<RemoteMediaWindow>;
   getSubtitleContextLines?: (range: { startTime: number; endTime: number }) => {
     previous: MediaTimingReviewContextLine[];
     next: MediaTimingReviewContextLine[];
   };
   decisionTimeoutMs?: number;
   openModal: (payload: MediaTimingReviewOpenPayload) => Promise<boolean>;
+  /** Tells the modal that the hidden player finished the previewed clip. */
+  onPreviewEnded?: (reviewId: string) => void;
   showStatus: (message: string) => void;
 }
 
@@ -124,6 +159,15 @@ export function collectMediaTimingContextLines(options: {
   return { previous, next };
 }
 
+/**
+ * Result for requests that name a review main has already resolved or disposed (decision
+ * watchdog, overlay teardown, duplicate modal). The renderer closes on it instead of
+ * leaving the user with controls that can never succeed.
+ */
+function staleReviewResult(): MediaTimingReviewActionResult {
+  return { ok: false, stale: true, message: 'This timing review is no longer active.' };
+}
+
 function isValidMediaTimingRange(
   payload: MediaTimingReviewOpenPayload,
   startTime: number,
@@ -199,6 +243,89 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
     }
   }
 
+  function ensureWindow(
+    review: ActiveReview,
+    range: RemoteMediaWindowRange,
+  ): Promise<RemoteMediaWindow | null> {
+    const { windowSource } = review;
+    if (!windowSource || review.windowFailed || !deps.acquireMediaWindow) {
+      return Promise.resolve(null);
+    }
+    const coversRange = (candidate: RemoteMediaWindowRange): boolean =>
+      candidate.startTime <= range.startTime && candidate.endTime >= range.endTime;
+    if (review.window && coversRange(review.window)) return Promise.resolve(review.window);
+    const inFlight = review.windowRequest;
+    if (inFlight && coversRange(inFlight)) return inFlight.promise;
+
+    const request = {
+      startTime: range.startTime,
+      endTime: range.endTime,
+      promise: Promise.resolve<RemoteMediaWindow | null>(null),
+    };
+    request.promise = deps
+      .acquireMediaWindow(windowSource, { startTime: range.startTime, endTime: range.endTime })
+      .then((window) => {
+        review.window = window;
+        return window;
+      })
+      .catch(() => {
+        // Fall back to the remote source for the rest of this review instead of retrying.
+        review.windowFailed = true;
+        return null;
+      })
+      .finally(() => {
+        if (review.windowRequest === request) review.windowRequest = null;
+      });
+    review.windowRequest = request;
+    return request.promise;
+  }
+
+  /**
+   * Returns the preview player for the range, restarting it when the range needs a
+   * different file (the first cached window, or a wider one after the timeline grew).
+   */
+  async function previewFor(
+    review: ActiveReview,
+    range: RemoteMediaWindowRange,
+  ): Promise<PreviewSession> {
+    const window = await ensureWindow(review, range);
+    if (active !== review) {
+      // The review ended during the download; do not start a player nobody will dispose.
+      throw new Error('This timing review is no longer active.');
+    }
+    const mediaPath = window?.path ?? review.mediaPath;
+    if (review.preview?.path === mediaPath) return review.preview.session;
+
+    const previous = review.preview;
+    const session = deps.createPreviewSession();
+    session.onPlaybackEnded(() => {
+      if (active === review && review.preview?.session === started) {
+        deps.onPreviewEnded?.(review.payload.reviewId);
+      }
+    });
+    const { audioTrackId, ...previewOptions } = review.previewOptions;
+    const started = session
+      .start({
+        mediaPath,
+        ...previewOptions,
+        // A cached window keeps one audio stream, so mpv's track id from the source no longer applies.
+        ...(window
+          ? { absoluteTimestamps: true }
+          : audioTrackId !== undefined
+            ? { audioTrackId }
+            : {}),
+      })
+      .then(() => session)
+      .catch((error) => {
+        session.dispose();
+        throw error;
+      });
+    review.preview = { path: mediaPath, session: started };
+    void started.catch(() => {});
+    if (previous) void previous.session.then((old) => old.dispose()).catch(() => {});
+    return started;
+  }
+
   async function runReview(request: MediaTimingReviewRequest): Promise<MediaTimingReviewDecision> {
     const mpvClient = deps.getMpvClient();
     const mediaPath =
@@ -208,11 +335,12 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
       return { action: 'use-original' };
     }
 
-    const [pauseRaw, durationRaw, audioTrackRaw, volumeRaw] = await Promise.all([
+    const [pauseRaw, durationRaw, audioTrackRaw, volumeRaw, resolvedSource] = await Promise.all([
       mpvClient.requestProperty?.('pause').catch(() => null) ?? null,
       mpvClient.requestProperty?.('duration').catch(() => null) ?? null,
       mpvClient.requestProperty?.('aid').catch(() => null) ?? null,
       mpvClient.requestProperty?.('volume').catch(() => null) ?? null,
+      deps.resolveMediaSource?.().catch(() => null) ?? null,
     ]);
     const pauseState = booleanProperty(pauseRaw);
     mpvClient.send({ command: ['set_property', 'pause', 'yes'] });
@@ -232,38 +360,51 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
       mediaDuration: finiteNumber(durationRaw) ?? undefined,
       ...(contextLines ? { contextLines } : {}),
     });
-    const previewSession = deps.createPreviewSession();
-    const preview = previewSession
-      .start({
-        mediaPath,
-        executablePath: deps.getMpvExecutablePath(),
-        audioTrackId: finiteNumber(audioTrackRaw) ?? undefined,
-        volume: finiteNumber(volumeRaw) ?? undefined,
-      })
-      .then(() => previewSession)
-      .catch((error) => {
-        previewSession.dispose();
-        throw error;
-      });
-    void preview.catch(() => {});
+    const sourcePath = resolvedSource?.path.trim() || mediaPath;
+    const inputOptions = resolvedSource?.inputOptions;
+    const audioStreamIndex =
+      resolvedSource?.singleResolvedStream || mpvClient.currentAudioStreamIndex == null
+        ? undefined
+        : mpvClient.currentAudioStreamIndex;
+    const windowSource: RemoteMediaWindowSource | null =
+      deps.acquireMediaWindow && isRemoteMediaWindowSourcePath(sourcePath)
+        ? {
+            path: sourcePath,
+            ...(inputOptions ? { inputOptions } : {}),
+            audioStreamIndex: audioStreamIndex ?? null,
+          }
+        : null;
 
     let resolveDecision!: (decision: MediaTimingReviewDecision) => void;
     const decisionPromise = new Promise<MediaTimingReviewDecision>((resolve) => {
       resolveDecision = resolve;
     });
-    active = {
+    const review: ActiveReview = {
       payload,
       mediaPath,
-      ...(mpvClient.currentAudioStreamIndex !== null &&
-      mpvClient.currentAudioStreamIndex !== undefined
-        ? { audioStreamIndex: mpvClient.currentAudioStreamIndex }
-        : {}),
+      waveformMedia: inputOptions ? { path: sourcePath, inputOptions } : sourcePath,
+      ...(audioStreamIndex !== undefined ? { audioStreamIndex } : {}),
+      windowSource,
+      window: null,
+      windowRequest: null,
+      windowFailed: false,
+      previewOptions: {
+        executablePath: deps.getMpvExecutablePath(),
+        audioTrackId: finiteNumber(audioTrackRaw) ?? undefined,
+        volume: finiteNumber(volumeRaw) ?? undefined,
+      },
+      preview: null,
       mpvClient,
       restorePlayback: pendingPauseRestore === mpvClient,
-      preview,
       resolve: resolveDecision,
     };
+    active = review;
     pendingPauseRestore = null;
+    // Download the visible timeline once now; the waveform and preview both wait on it.
+    void previewFor(review, {
+      startTime: payload.timelineStartTime,
+      endTime: payload.timelineEndTime,
+    }).catch(() => {});
 
     const opened = await deps.openModal(payload).catch(() => false);
     if (!opened) {
@@ -311,16 +452,27 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
   ): Promise<MediaTimingReviewActionResult> {
     const current = active;
     if (!current || request.reviewId !== current.payload.reviewId) {
-      return { ok: false, message: 'This timing review is no longer active.' };
+      return staleReviewResult();
     }
     if (!isValidMediaTimingRange(current.payload, request.startTime, request.endTime)) {
       return { ok: false, message: 'The selected preview range is invalid.' };
     }
     try {
-      const previewSession = await current.preview;
+      const previewSession = await previewFor(current, request);
+      if (active !== current) {
+        return staleReviewResult();
+      }
       await previewSession.play(request.startTime, request.endTime);
+      // Playback spans the whole clip, so the review can end (watchdog, teardown) while
+      // it runs; reporting success would leave the modal open on a dead review.
+      if (active !== current) {
+        return staleReviewResult();
+      }
       return { ok: true };
     } catch (error) {
+      if (active !== current) {
+        return staleReviewResult();
+      }
       return {
         ok: false,
         message: `Audio preview unavailable: ${error instanceof Error ? error.message : String(error)}`,
@@ -333,7 +485,7 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
   ): Promise<MediaTimingReviewWaveformResult> {
     const current = active;
     if (!current || request.reviewId !== current.payload.reviewId) {
-      return { ok: false, message: 'This timing review is no longer active.' };
+      return staleReviewResult();
     }
     if (
       !Number.isFinite(request.startTime) ||
@@ -347,19 +499,30 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
     }
 
     try {
+      const window = await ensureWindow(current, request);
+      if (active !== current) {
+        return staleReviewResult();
+      }
       const peaks = await deps.generateWaveform({
-        mediaPath: current.mediaPath,
+        mediaPath: window?.media ?? current.waveformMedia,
         startTime: request.startTime,
         endTime: request.endTime,
-        ...(current.audioStreamIndex !== undefined
+        ...(!window && current.audioStreamIndex !== undefined
           ? { audioStreamIndex: current.audioStreamIndex }
           : {}),
       });
-      if (active !== current || peaks.length < 2 || peaks.some((peak) => !Number.isFinite(peak))) {
+      // ffmpeg decoding runs long enough for the review to end underneath it.
+      if (active !== current) {
+        return staleReviewResult();
+      }
+      if (peaks.length < 2 || peaks.some((peak) => !Number.isFinite(peak))) {
         return { ok: false, message: 'Timing waveform is unavailable.' };
       }
       return { ok: true, peaks };
     } catch {
+      if (active !== current) {
+        return staleReviewResult();
+      }
       return { ok: false, message: 'Timing waveform is unavailable.' };
     }
   }
@@ -367,11 +530,11 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
   async function stopPreview(reviewId: string): Promise<MediaTimingReviewActionResult> {
     const current = active;
     if (!current || reviewId !== current.payload.reviewId) {
-      return { ok: false, message: 'This timing review is no longer active.' };
+      return staleReviewResult();
     }
     try {
-      const previewSession = await current.preview;
-      await previewSession.stop();
+      const previewSession = current.preview ? await current.preview.session : null;
+      await previewSession?.stop();
       return { ok: true };
     } catch (error) {
       return {
@@ -384,7 +547,7 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
   function resolveReview(request: MediaTimingReviewResolveRequest): MediaTimingReviewActionResult {
     const current = active;
     if (!current || request.reviewId !== current.payload.reviewId) {
-      return { ok: false, message: 'This timing review is no longer active.' };
+      return staleReviewResult();
     }
     if (request.decision.action === 'confirm') {
       const { startTime, endTime, text } = request.decision;
@@ -403,7 +566,7 @@ export function createMediaTimingReviewRuntime(deps: MediaTimingReviewRuntimeDep
     const current = active;
     active = null;
     if (!current) return;
-    void current.preview.then((session) => session.dispose()).catch(() => {});
+    void current.preview?.session.then((session) => session.dispose()).catch(() => {});
     if (current.restorePlayback && current.mpvClient.connected) {
       current.mpvClient.send({ command: ['set_property', 'pause', 'no'] });
     }

src/main/runtime/mpv-main-event-main-deps.test.ts

@@ -1,5 +1,6 @@
 import assert from 'node:assert/strict';
 import test from 'node:test';
+import { parseAssCues } from '../../core/services/subtitle-cue-parser';
 import { createBuildBindMpvMainEventHandlersMainDepsHandler } from './mpv-main-event-main-deps';
 
 test('mpv main event main deps map app state updates and delegate callbacks', async () => {
@@ -503,14 +504,21 @@ test('canonical ASS cues replace live glyph spam for display, history, and immer
   assert.deepEqual(timing.slice(3), [{ text: '今　手にある物差しでは', start: 1.2, end: 3.8 }]);
   assert.equal(immersion.length, 3);
 
-  // A jump of exactly the seek threshold counts as a seek, matching the time-pos
-  // handler's own `>=` boundary.
-  handlers.onTimePosUpdate?.(4.5);
-  handlers.onTimePosUpdate?.(2);
+  // Jumping back to a brief previous line moves time-pos by less than the general
+  // seek threshold. It is still a backward seek, so the revisited line records
+  // again; otherwise multi-line copy would keep treating the later line as current.
+  handlers.onTimePosUpdate?.(3.9);
+  handlers.onTimePosUpdate?.(2.9);
   handlers.recordSubtitleTiming('今', 0.8, 1.5);
 
   assert.deepEqual(timing.slice(4), [{ text: '今　手にある物差しでは', start: 1.2, end: 3.8 }]);
 
+  // Tiny time-pos jitter is not a seek and must not re-record the line.
+  handlers.onTimePosUpdate?.(3.0);
+  handlers.onTimePosUpdate?.(2.9);
+  handlers.recordSubtitleTiming('今', 0.8, 1.5);
+  assert.equal(timing.length, 5);
+
   handlers.recordImmersionSubtitleLine('Maid\nCafe', 10, 12);
   handlers.recordSubtitleTiming('Maid\nCafe', 10, 12);
   assert.equal(immersion.length, 3);
@@ -575,3 +583,156 @@ test('subtitle-track changes stop stale canonical cues from substituting immedia
   assert.equal(appState.activeParsedSubtitleSource, null);
   assert.equal(handlers.resolveSubtitleText?.('今\n手にある'), '今\n手にある');
 });
+
+test('subtitle recorders drop ASS furigana events the same way the display does', () => {
+  // Broadcast-caption ASS (Caption2Ass style): furigana are separate half-scale events
+  // positioned above their base line, and mpv lists them as their own live lines.
+  const cues = parseAssCues(
+    [
+      '[Script Info]',
+      'PlayResX: 960',
+      'PlayResY: 540',
+      '',
+      '[V4+ Styles]',
+      'Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding',
+      'Style: Default,Yu Gothic,46,&H00FFFFFF,&H000000FF,&H00000000,&H7F000000,1,0,0,0,100,100,4,0,1,2,2,1,0,0,0,1',
+      '',
+      '[Events]',
+      'Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text',
+      'Dialogue: 0,0:04:56.26,0:04:59.63,Default,,0000,0000,0000,,{\\pos(472,443)\\fscx50\\fscy50}あくむ',
+      'Dialogue: 0,0:04:56.26,0:04:59.63,Default,,0000,0000,0000,,{\\pos(172,497)}こんな短時間で{\\fscx50}　{\\fscx100}悪夢{\\fscx50}　{\\fscx100}見んなよ{\\fscx50}。',
+      'Dialogue: 0,0:04:59.63,0:05:03.54,Default,,0000,0000,0000,,{\\pos(172,407)\\fscx50}（{\\fscx100}平{\\fscx50}）{\\fscx100}暗記教科は{\\fscx50}　{\\fscx100}もう',
+      'Dialogue: 0,0:04:59.63,0:05:03.54,Default,,0000,0000,0000,,{\\pos(332,443)\\fscx50\\fscy50}かた',
+      'Dialogue: 0,0:04:59.63,0:05:03.54,Default,,0000,0000,0000,,{\\pos(412,443)\\fscx50\\fscy50}ぱし',
+      'Dialogue: 0,0:04:59.63,0:05:03.54,Default,,0000,0000,0000,,{\\pos(172,497)}とにかく片っ端から覚えるんだよ{\\fscx50}。',
+    ].join('\n'),
+  );
+  assert.deepEqual(
+    cues.map((cue) => cue.text),
+    [
+      'こんな短時間で　悪夢　見んなよ。',
+      '（平）暗記教科は　もう',
+      'とにかく片っ端から覚えるんだよ。',
+    ],
+  );
+
+  const immersion: string[] = [];
+  const timing: string[] = [];
+  const handlers = createBuildBindMpvMainEventHandlersMainDepsHandler({
+    appState: {
+      initialArgs: null,
+      overlayRuntimeInitialized: true,
+      mpvClient: { currentTimePos: 299.7 },
+      immersionTracker: { recordSubtitleLine: (text: string) => immersion.push(text) },
+      subtitleTimingTracker: { recordSubtitle: (text: string) => timing.push(text) },
+      activeParsedSubtitleCues: cues,
+      currentSubText: '',
+      currentSubAssText: '',
+      playbackPaused: null,
+      previousSecondarySubVisibility: false,
+    },
+    getQuitOnDisconnectArmed: () => false,
+    scheduleQuitCheck: () => {},
+    quitApp: () => {},
+    reportJellyfinRemoteStopped: () => {},
+    syncOverlayMpvSubtitleSuppression: () => {},
+    maybeRunAnilistPostWatchUpdate: async () => {},
+    logSubtitleTimingError: () => {},
+    broadcastToOverlayWindows: () => {},
+    onSubtitleChange: () => {},
+    ensureImmersionTrackerInitialized: () => {},
+    updateCurrentMediaPath: () => {},
+    restoreMpvSubVisibility: () => {},
+    getCurrentAnilistMediaKey: () => null,
+    resetAnilistMediaTracking: () => {},
+    maybeProbeAnilistDuration: () => {},
+    ensureAnilistMediaGuess: () => {},
+    syncImmersionMediaState: () => {},
+    updateCurrentMediaTitle: () => {},
+    resetAnilistMediaGuessState: () => {},
+    reportJellyfinRemoteProgress: () => {},
+    updateSubtitleRenderMetrics: () => {},
+    refreshDiscordPresence: () => {},
+  })();
+
+  const liveText = '（平）暗記教科は　もう\nかた\nぱし\nとにかく片っ端から覚えるんだよ。';
+  const expected = '（平）暗記教科は　もう\n\nとにかく片っ端から覚えるんだよ。';
+  assert.equal(handlers.resolveSubtitleText?.(liveText), expected);
+  handlers.recordImmersionSubtitleLine(liveText, 299.63, 303.54);
+  handlers.recordSubtitleTiming(liveText, 299.63, 303.54);
+
+  assert.deepEqual(immersion, [expected]);
+  assert.deepEqual(timing, [expected]);
+});
+
+test('a resolved line survives recording while a fragment grid is on screen', () => {
+  // Fragment stripping drops every line it can trace back to a cue, and returns nothing
+  // at all when a fragment grid is nearby. Text the parsed cues already resolved is a
+  // complete line, not raw mpv output, so it must not be fed through that path.
+  const immersion: string[] = [];
+  const timing: string[] = [];
+  const handlers = createBuildBindMpvMainEventHandlersMainDepsHandler({
+    appState: {
+      initialArgs: null,
+      overlayRuntimeInitialized: true,
+      mpvClient: { currentTimePos: 3.2 },
+      immersionTracker: { recordSubtitleLine: (text: string) => immersion.push(text) },
+      subtitleTimingTracker: { recordSubtitle: (text: string) => timing.push(text) },
+      activeParsedSubtitleCues: [
+        { startTime: 3, endTime: 6, text: '飛び越えてみたくて', source: 'canonical-ass' },
+        {
+          startTime: 3,
+          endTime: 6,
+          text: 'MaidCafeMaidCafe',
+          source: 'reconstructed-ass',
+          assLayout: { kind: 'fragment-grid', sourceOrder: 2 },
+        },
+      ],
+      currentSubText: '',
+      currentSubAssText: '',
+      playbackPaused: null,
+      previousSecondarySubVisibility: false,
+    },
+    getQuitOnDisconnectArmed: () => false,
+    scheduleQuitCheck: () => {},
+    quitApp: () => {},
+    reportJellyfinRemoteStopped: () => {},
+    syncOverlayMpvSubtitleSuppression: () => {},
+    maybeRunAnilistPostWatchUpdate: async () => {},
+    logSubtitleTimingError: () => {},
+    broadcastToOverlayWindows: () => {},
+    onSubtitleChange: () => {},
+    ensureImmersionTrackerInitialized: () => {},
+    updateCurrentMediaPath: () => {},
+    restoreMpvSubVisibility: () => {},
+    getCurrentAnilistMediaKey: () => null,
+    resetAnilistMediaTracking: () => {},
+    maybeProbeAnilistDuration: () => {},
+    ensureAnilistMediaGuess: () => {},
+    syncImmersionMediaState: () => {},
+    updateCurrentMediaTitle: () => {},
+    resetAnilistMediaGuessState: () => {},
+    reportJellyfinRemoteProgress: () => {},
+    updateSubtitleRenderMetrics: () => {},
+    refreshDiscordPresence: () => {},
+  })();
+
+  // The grid fragment beside the lyric keeps canonical substitution from applying, so
+  // recording falls to the parsed view -- which is where the whole line is recovered.
+  const liveText = '飛び越え\nMaid';
+  assert.equal(handlers.resolveSubtitleText?.(liveText), '飛び越えてみたくて');
+  handlers.recordImmersionSubtitleLine(liveText, 3, 6);
+  handlers.recordSubtitleTiming(liveText, 3, 6);
+
+  assert.deepEqual(immersion, ['飛び越えてみたくて']);
+  assert.deepEqual(timing, ['飛び越えてみたくて']);
+
+  // A spacer event left as literal control debris is not a subtitle line. The display
+  // drops it, so no recorder may keep it either.
+  assert.equal(handlers.resolveSubtitleText?.('\\'), '');
+  handlers.recordImmersionSubtitleLine('\\', 3, 6);
+  handlers.recordSubtitleTiming('\\', 3, 6);
+
+  assert.equal(immersion.length, 1);
+  assert.equal(timing.length, 1);
+});

src/main/runtime/mpv-main-event-main-deps.ts

@@ -1,10 +1,9 @@
 import { createSubtitleLineDedupGate } from '../../core/services/subtitle-line-dedup-gate';
 import type { MergedToken, SubtitleCue, SubtitleData } from '../../types';
-import { SEEK_LIKE_TIME_DELTA_SECONDS } from './mpv-main-event-actions';
 import {
   resolveCanonicalPrimarySubtitle,
   resolvePrimarySubtitleText,
-  stripCanonicalFragmentLines,
+  resolveRecordedPrimarySubtitleText,
 } from './primary-subtitle-text';
 
 type AnilistPostWatchRunOptions = {
@@ -115,6 +114,8 @@ export function createBuildBindMpvMainEventHandlersMainDepsHandler(deps: {
   // after the change is dropped instead of landing in the next session.
   let subtitleSessionEpoch = 0;
   let lastTimePosForTimingReset: number | null = null;
+  // Small margin so time-pos jitter is not mistaken for a backward seek.
+  const BACKWARD_SEEK_TIMING_RESET_SECONDS = 0.25;
   const canonicalCueKey = (cue: SubtitleCue): string =>
     `${cue.startTime}|${cue.endTime}|${cue.text}`;
   const resetSubtitleDeduplication = (): void => {
@@ -130,10 +131,12 @@ export function createBuildBindMpvMainEventHandlersMainDepsHandler(deps: {
       currentTimeSec: startSec,
       cues: deps.appState.activeParsedSubtitleCues,
     });
-  // When substitution declined because dialogue shares the screen with a song, record
-  // the dialogue alone rather than the combined dialogue-plus-fragments stack.
-  const stripFragmentsForRecording = (liveText: string, startSec: number) =>
-    stripCanonicalFragmentLines({
+  // Recorders see the same text the overlay displays: mpv's live `sub-text` lists every
+  // simultaneously active ASS event, including furigana events the parser folded into
+  // their base line. Cues the caller already resolved canonically are recorded one by
+  // one above; everything else goes through the shared recording resolution.
+  const resolveTextForRecording = (liveText: string, startSec: number): string =>
+    resolveRecordedPrimarySubtitleText({
       liveText,
       currentTimeSec: startSec,
       cues: deps.appState.activeParsedSubtitleCues,
@@ -217,7 +220,7 @@ export function createBuildBindMpvMainEventHandlersMainDepsHandler(deps: {
         }
         return;
       }
-      text = stripFragmentsForRecording(text, start);
+      text = resolveTextForRecording(text, start);
       if (!text.trim()) {
         return;
       }
@@ -231,7 +234,7 @@ export function createBuildBindMpvMainEventHandlersMainDepsHandler(deps: {
       const secondaryText = deps.appState.mpvClient?.currentSecondarySubText || undefined;
       const canonical = resolveCanonicalSample(text, start);
       if (!canonical) {
-        const recordableText = stripFragmentsForRecording(text, start);
+        const recordableText = resolveTextForRecording(text, start);
         if (!recordableText.trim()) {
           return;
         }
@@ -344,13 +347,15 @@ export function createBuildBindMpvMainEventHandlersMainDepsHandler(deps: {
       deps.reportJellyfinRemoteProgress(forceImmediate),
     consumeExplicitSeek: deps.consumeExplicitSeek,
     onTimePosUpdate: (time: number) => {
-      // Timing history is a viewing log: after a real backward seek, a rewatched
-      // canonical line should enter it again. Immersion stats keep their
-      // once-per-media deduplication and are not reset here.
+      // Timing history is a viewing log: after any backward seek, a rewatched canonical
+      // line should enter it again so multi-line copy treats it as the current line.
+      // Playback never moves time-pos backward on its own, so even a short jump to a
+      // brief previous line counts. Immersion stats keep their once-per-media
+      // deduplication and are not reset here.
       if (
         Number.isFinite(time) &&
         lastTimePosForTimingReset !== null &&
-        time <= lastTimePosForTimingReset - SEEK_LIKE_TIME_DELTA_SECONDS
+        time <= lastTimePosForTimingReset - BACKWARD_SEEK_TIMING_RESET_SECONDS
       ) {
         recordedTimingCanonicalKeys.clear();
       }

src/main/runtime/primary-subtitle-text.test.ts

@@ -3,6 +3,7 @@ import test from 'node:test';
 import { parseSubtitleCues } from '../../core/services/subtitle-cue-parser';
 import {
   resolveCanonicalPrimarySubtitle,
+  resolvePrimarySubtitle,
   resolvePrimarySubtitleText,
   stripCanonicalFragmentLines,
 } from './primary-subtitle-text';
@@ -674,3 +675,65 @@ test('resolvePrimarySubtitleText keeps source order when no cue declares a place
     'First line\n\nSecond line',
   );
 });
+
+test('resolvePrimarySubtitleText publishes a wrapped caption sentence as one cue', () => {
+  // mpv still reports the two source rows (plus the ruby row) as separate live lines, so
+  // the merged cue must explain all of them and come back as a single-break line the
+  // display layer may flatten, not as a two-cue boundary.
+  const ass = [
+    '[Script Info]',
+    'PlayResY: 540',
+    '',
+    '[Events]',
+    'Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text',
+    'Dialogue: 0,0:02:42.33,0:02:44.43,Default,,0,0,0,,{\\pos(232,437)\\fscx50}（{\\fscx100}東{\\fscx50}）{\\fscx100}≪好きだと',
+    'Dialogue: 0,0:02:42.33,0:02:44.43,Default,,0,0,0,,{\\pos(292,443)\\fscx50\\fscy50}じかく',
+    'Dialogue: 0,0:02:42.33,0:02:44.43,Default,,0,0,0,,{\\pos(232,497)}自覚してしまったものの➡',
+    // A second labeled turn, so the script reads as broadcast captions.
+    'Dialogue: 0,0:02:44.43,0:02:47.37,Default,,0,0,0,,{\\pos(212,497)\\fscx50}（{\\fscx100}平{\\fscx50}）{\\fscx100}どうした？',
+  ].join('\n');
+  const cues = parseSubtitleCues(ass, 'polar-opposites-s02e09.ass');
+
+  assert.equal(
+    resolvePrimarySubtitleText({
+      liveText: '（東）≪好きだと\nじかく\n自覚してしまったものの➡',
+      currentTimeSec: 163,
+      cues,
+    }),
+    '（東）≪好きだと\n自覚してしまったものの➡',
+  );
+});
+
+test('resolvePrimarySubtitle drops a finished caption row lingering beside a fresh line', () => {
+  // Broadcast captions give each row its own event, and a row of the previous line can
+  // outlive its siblings by a frame. mpv's sub-text still lists it, so the mined line
+  // must come from the parsed cue that is actually running, with that cue's timings.
+  const ass = [
+    '[Script Info]',
+    'PlayResY: 540',
+    '',
+    '[Events]',
+    'Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text',
+    'Dialogue: 0,0:14:30.00,0:14:33.00,Default,,0,0,0,,{\\pos(232,437)\\fscx50}（{\\fscx100}東{\\fscx50}）{\\fscx100}ずっと　言えなかっ',
+    'Dialogue: 0,0:14:30.00,0:14:33.02,Default,,0,0,0,,{\\pos(232,497)}たが',
+    'Dialogue: 0,0:14:33.00,0:14:36.00,Default,,0,0,0,,{\\pos(232,437)}⸨もし　お互い',
+    'Dialogue: 0,0:14:33.00,0:14:36.00,Default,,0,0,0,,{\\pos(232,497)}本命　受かったら　大学　近いし➡',
+  ].join('\n');
+  const cues = parseSubtitleCues(ass, 'polar-opposites-s02e09.ass');
+
+  const resolved = resolvePrimarySubtitle({
+    liveText: 'たが\n⸨もし　お互い\n本命　受かったら　大学　近いし➡',
+    currentTimeSec: 14 * 60 + 33.05,
+    cues,
+  });
+
+  assert.deepEqual(
+    { ...resolved, cues: resolved?.cues.map((cue) => cue.text) },
+    {
+      text: '⸨もし　お互い\n本命　受かったら　大学　近いし➡',
+      startTime: 14 * 60 + 33,
+      endTime: 14 * 60 + 36,
+      cues: ['⸨もし　お互い\n本命　受かったら　大学　近いし➡'],
+    },
+  );
+});

src/main/runtime/primary-subtitle-text.ts

@@ -26,6 +26,13 @@ function cuesUseAssSyntax(cues: readonly SubtitleCue[] | null | undefined): bool
   );
 }
 
+function decodedLiveText(
+  liveText: string,
+  cues: readonly SubtitleCue[] | null | undefined,
+): string {
+  return cuesUseAssSyntax(cues) ? removeAssControlDebrisLines(liveText) : liveText;
+}
+
 function animationSpan(cue: SubtitleCue): { start: number; end: number } {
   return {
     start: cue.animationStartTime ?? cue.startTime,
@@ -290,24 +297,56 @@ export function stripCanonicalFragmentLines(options: {
   return removeLiveGlyphFragmentLines(options.liveText);
 }
 
+/**
+ * Recording text for a live sample. Callers substitute canonical cues themselves and
+ * record those cue by cue, so what is resolved here is the parsed view -- the one that
+ * folds ASS furigana events back into their base line. Its text is already a complete
+ * line, while fragment stripping takes raw mpv text and would discard a resolved line
+ * whole while a fragment grid is on screen, so only one of the two ever runs.
+ */
+export function resolveRecordedPrimarySubtitleText(options: {
+  liveText: string;
+  currentTimeSec: number;
+  cues: readonly SubtitleCue[] | null | undefined;
+}): string {
+  const liveText = decodedLiveText(options.liveText, options.cues);
+  if (!liveText.trim()) {
+    return liveText;
+  }
+  return (
+    resolveActiveParsedPrimarySubtitle({ ...options, liveText })?.text ??
+    stripCanonicalFragmentLines({ ...options, liveText })
+  );
+}
+
+/**
+ * The parsed view of the live text with its cue timings: a canonical animation when one
+ * explains the live lines, otherwise the active parsed cues. Null when the parsed cues
+ * cannot account for every live line, in which case callers keep the raw mpv text.
+ */
+export function resolvePrimarySubtitle(options: {
+  liveText: string;
+  currentTimeSec: number;
+  cues: readonly SubtitleCue[] | null | undefined;
+}): ResolvedPrimarySubtitle | null {
+  const liveText = decodedLiveText(options.liveText, options.cues);
+  if (!liveText.trim()) {
+    return null;
+  }
+  return (
+    resolveCanonicalPrimarySubtitle({ ...options, liveText }) ??
+    resolveActiveParsedPrimarySubtitle({ ...options, liveText })
+  );
+}
+
 export function resolvePrimarySubtitleText(options: {
   liveText: string;
   currentTimeSec: number;
   cues: readonly SubtitleCue[] | null | undefined;
 }): string {
-  const liveText = cuesUseAssSyntax(options.cues)
-    ? removeAssControlDebrisLines(options.liveText)
-    : options.liveText;
+  const liveText = decodedLiveText(options.liveText, options.cues);
   if (!liveText.trim()) {
     return liveText;
   }
-  return (
-    resolveCanonicalPrimarySubtitle({
-      liveText,
-      currentTimeSec: options.currentTimeSec,
-      cues: options.cues,
-    })?.text ??
-    resolveActiveParsedPrimarySubtitle({ ...options, liveText })?.text ??
-    removeLiveGlyphFragmentLines(liveText)
-  );
+  return resolvePrimarySubtitle(options)?.text ?? removeLiveGlyphFragmentLines(liveText);
 }

src/media-generator.test.ts

@@ -10,6 +10,9 @@ import {
   MediaGenerator,
   type MediaGeneratorOptions,
 } from './media-generator';
+import { RemoteMediaWindowCache } from './core/services/remote-media-window-cache';
+
+const REMOTE_STREAM_URL = 'https://jellyfin.example/Videos/abc/stream?static=true';
 
 async function withStubbedFfmpeg(
   run: (generator: MediaGenerator, argsPath: string) => Promise<void>,
@@ -21,6 +24,7 @@ async function withStubbedFfmpeg(
   const root = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-media-generator-test-'));
   const binDir = path.join(root, 'bin');
   const tempDir = path.join(root, 'media');
+  const windowsDir = path.join(root, 'windows');
   const argsPath = path.join(root, 'ffmpeg-args.txt');
   fs.mkdirSync(binDir, { recursive: true });
   const ffmpegStubPath = path.join(binDir, 'ffmpeg-stub.cjs');
@@ -34,7 +38,7 @@ async function withStubbedFfmpeg(
       "  console.log(' V..... libaom-av1');",
       '  process.exit(0);',
       '}',
-      "fs.writeFileSync(process.env.SUBMINER_TEST_FFMPEG_ARGS, JSON.stringify(args), 'utf8');",
+      "fs.appendFileSync(process.env.SUBMINER_TEST_FFMPEG_ARGS, JSON.stringify(args) + '\\n', 'utf8');",
       'const outputPath = args.at(-1);',
       "if (process.env.SUBMINER_TEST_FFMPEG_SKIP_OUTPUT !== '1') {",
       "  fs.writeFileSync(outputPath, 'avif', 'utf8');",
@@ -61,12 +65,18 @@ async function withStubbedFfmpeg(
   } else {
     delete process.env.SUBMINER_TEST_FFMPEG_SKIP_OUTPUT;
   }
-  const generator = new MediaGenerator(tempDir, options);
+  // Each test gets its own window cache so remote inputs never leak windows between tests.
+  const remoteMediaWindows = new RemoteMediaWindowCache({ tempDir: windowsDir, idleTtlMs: 0 });
+  const generator = new MediaGenerator(tempDir, {
+    remoteMediaWindows,
+    ...options,
+  });
 
   try {
     await run(generator, argsPath);
   } finally {
     generator.cleanup();
+    remoteMediaWindows.cleanup();
     process.env.PATH = originalPath;
     if (originalArgsPath === undefined) {
       delete process.env.SUBMINER_TEST_FFMPEG_ARGS;
@@ -82,8 +92,17 @@ async function withStubbedFfmpeg(
   }
 }
 
+function readAllFfmpegArgs(argsPath: string): string[][] {
+  return fs
+    .readFileSync(argsPath, 'utf8')
+    .split('\n')
+    .filter((line) => line.trim().length > 0)
+    .map((line) => JSON.parse(line) as string[]);
+}
+
+/** Arguments of the most recent ffmpeg invocation. */
 function readFfmpegArgs(argsPath: string): string[] {
-  return JSON.parse(fs.readFileSync(argsPath, 'utf8')) as string[];
+  return readAllFfmpegArgs(argsPath).at(-1) ?? [];
 }
 
 test('buildAnimatedImageVideoFilter holds lead-in until the next frame after the audio boundary', () => {
@@ -272,41 +291,131 @@ test('generateAudio recreates missing temp directory before invoking ffmpeg', as
 });
 
 test('generateAudio adds remote input options before the ffmpeg input', async () => {
-  await withStubbedFfmpeg(async (generator, argsPath) => {
-    await generator.generateAudio(
-      {
-        path: 'https://rr1---sn.example.googlevideo.com/videoplayback?mime=audio%2Fwebm',
-        inputOptions: {
-          reconnect: true,
-          userAgent: 'Mozilla/5.0',
-          headers: {
-            Referer: 'https://www.youtube.com/',
-            Origin: 'https://www.youtube.com',
+  await withStubbedFfmpeg(
+    async (generator, argsPath) => {
+      await generator.generateAudio(
+        {
+          path: 'https://rr1---sn.example.googlevideo.com/videoplayback?mime=audio%2Fwebm',
+          inputOptions: {
+            reconnect: true,
+            userAgent: 'Mozilla/5.0',
+            headers: {
+              Referer: 'https://www.youtube.com/',
+              Origin: 'https://www.youtube.com',
+            },
           },
         },
-      },
+        10,
+        12,
+      );
+
+      const args = readFfmpegArgs(argsPath);
+      const inputIndex = args.indexOf('-i');
+      assert.ok(inputIndex > 0);
+      assert.ok(args.indexOf('-reconnect') > -1);
+      assert.ok(args.indexOf('-reconnect') < inputIndex);
+      assert.equal(args[args.indexOf('-reconnect') + 1], '1');
+      assert.equal(args[args.indexOf('-reconnect_streamed') + 1], '1');
+      assert.equal(args[args.indexOf('-reconnect_on_network_error') + 1], '1');
+      assert.equal(args[args.indexOf('-reconnect_on_http_error') + 1], '403,5xx');
+      assert.equal(args[args.indexOf('-reconnect_delay_max') + 1], '5');
+      assert.equal(args[args.indexOf('-user_agent') + 1], 'Mozilla/5.0');
+      assert.equal(
+        args[args.indexOf('-headers') + 1],
+        'Referer: https://www.youtube.com/\r\nOrigin: https://www.youtube.com\r\n',
+      );
+    },
+    { remoteMediaWindows: null },
+  );
+});
+
+test('generateAudio downloads a remote window once and extracts from it with absolute seeks', async () => {
+  await withStubbedFfmpeg(async (generator, argsPath) => {
+    await generator.generateAudio(
+      { path: REMOTE_STREAM_URL, inputOptions: { reconnect: true } },
       10,
       12,
+      0.5,
+      2,
     );
 
-    const args = readFfmpegArgs(argsPath);
-    const inputIndex = args.indexOf('-i');
-    assert.ok(inputIndex > 0);
-    assert.ok(args.indexOf('-reconnect') > -1);
-    assert.ok(args.indexOf('-reconnect') < inputIndex);
-    assert.equal(args[args.indexOf('-reconnect') + 1], '1');
-    assert.equal(args[args.indexOf('-reconnect_streamed') + 1], '1');
-    assert.equal(args[args.indexOf('-reconnect_on_network_error') + 1], '1');
-    assert.equal(args[args.indexOf('-reconnect_on_http_error') + 1], '403,5xx');
-    assert.equal(args[args.indexOf('-reconnect_delay_max') + 1], '5');
-    assert.equal(args[args.indexOf('-user_agent') + 1], 'Mozilla/5.0');
-    assert.equal(
-      args[args.indexOf('-headers') + 1],
-      'Referer: https://www.youtube.com/\r\nOrigin: https://www.youtube.com\r\n',
-    );
+    const calls = readAllFfmpegArgs(argsPath);
+    assert.equal(calls.length, 2);
+    const [fetchArgs, audioArgs] = calls as [string[], string[]];
+    assert.equal(fetchArgs[fetchArgs.indexOf('-i') + 1], REMOTE_STREAM_URL);
+    assert.ok(fetchArgs.indexOf('-reconnect') < fetchArgs.indexOf('-i'));
+    assert.equal(fetchArgs[fetchArgs.indexOf('-ss') + 1], '9.25');
+    assert.equal(fetchArgs[fetchArgs.lastIndexOf('-map') + 1], '0:2');
+    assert.ok(fetchArgs.includes('-copyts'));
+
+    const windowPath = audioArgs[audioArgs.indexOf('-i') + 1];
+    assert.ok(windowPath?.endsWith('.mkv'));
+    assert.notEqual(windowPath, REMOTE_STREAM_URL);
+    assert.equal(audioArgs[audioArgs.indexOf('-ss') + 1], '9.5');
+    assert.equal(audioArgs[audioArgs.indexOf('-seek_timestamp') + 1], '1');
+    assert.ok(audioArgs.indexOf('-seek_timestamp') < audioArgs.indexOf('-i'));
+    assert.equal(audioArgs.includes('-reconnect'), false);
+    assert.equal(audioArgs.includes('-map'), false);
+    assert.equal(audioArgs.includes('-probesize'), false);
+    assert.ok(audioArgs.includes('loudnorm=I=-23:TP=-2:LRA=11'));
   });
 });
 
+test('generateScreenshot reuses a downloaded window but never downloads one itself', async () => {
+  await withStubbedFfmpeg(async (generator, argsPath) => {
+    await generator.generateScreenshot(REMOTE_STREAM_URL, 11, { format: 'jpg' });
+    let calls = readAllFfmpegArgs(argsPath);
+    assert.equal(calls.length, 1);
+    assert.equal(calls[0]![calls[0]!.indexOf('-i') + 1], REMOTE_STREAM_URL);
+
+    await generator.generateAudio(REMOTE_STREAM_URL, 10, 12);
+    await generator.generateScreenshot(REMOTE_STREAM_URL, 11, { format: 'jpg' });
+    await generator.generateScreenshot(REMOTE_STREAM_URL, 40, { format: 'jpg' });
+
+    calls = readAllFfmpegArgs(argsPath);
+    assert.equal(calls.length, 5);
+    const insideWindow = calls[3]!;
+    assert.ok(insideWindow[insideWindow.indexOf('-i') + 1]?.endsWith('.mkv'));
+    assert.equal(insideWindow[insideWindow.indexOf('-seek_timestamp') + 1], '1');
+    const outsideWindow = calls[4]!;
+    assert.equal(outsideWindow[outsideWindow.indexOf('-i') + 1], REMOTE_STREAM_URL);
+  });
+});
+
+test('generateAnimatedImage downloads the clip window before encoding', async () => {
+  await withStubbedFfmpeg(async (generator, argsPath) => {
+    await generator.generateAnimatedImage(REMOTE_STREAM_URL, 10, 12, 0, { fps: 10 });
+
+    const calls = readAllFfmpegArgs(argsPath).filter(
+      (args) => args[0] !== '-hide_banner' || args[1] !== '-encoders',
+    );
+    assert.equal(calls.length, 2);
+    assert.equal(calls[0]![calls[0]!.indexOf('-i') + 1], REMOTE_STREAM_URL);
+    assert.ok(calls[1]![calls[1]!.indexOf('-i') + 1]?.endsWith('.mkv'));
+    assert.equal(calls[1]![calls[1]!.indexOf('-seek_timestamp') + 1], '1');
+  });
+});
+
+test('generateAudio reads the remote source directly when the window download fails', async () => {
+  await withStubbedFfmpeg(
+    async (generator, argsPath) => {
+      await generator.generateAudio(REMOTE_STREAM_URL, 10, 12);
+
+      const args = readFfmpegArgs(argsPath);
+      assert.equal(args[args.indexOf('-i') + 1], REMOTE_STREAM_URL);
+      assert.equal(args.includes('-seek_timestamp'), false);
+    },
+    {
+      remoteMediaWindows: new RemoteMediaWindowCache({
+        execFile: (_file, _args, _options, callback) =>
+          queueMicrotask(() => callback(Object.assign(new Error('offline'), { code: 1 }))),
+        idleTtlMs: 0,
+        logDebug: () => undefined,
+      }),
+    },
+  );
+});
+
 test('generateAudio skips stale audio stream maps for single resolved streams', async () => {
   await withStubbedFfmpeg(async (generator, argsPath) => {
     await generator.generateAudio(
@@ -320,8 +429,9 @@ test('generateAudio skips stale audio stream maps for single resolved streams',
       22,
     );
 
-    const args = readFfmpegArgs(argsPath);
-    assert.equal(args.includes('-map'), false);
+    const [fetchArgs, audioArgs] = readAllFfmpegArgs(argsPath) as [string[], string[]];
+    assert.equal(fetchArgs[fetchArgs.lastIndexOf('-map') + 1], '0:a');
+    assert.equal(audioArgs.includes('-map'), false);
   });
 });
 

src/media-generator.ts

@@ -22,6 +22,12 @@ import * as path from 'path';
 import * as os from 'os';
 import { createLogger } from './logger';
 import { normalizeMediaInput, type MediaInput } from './media-input';
+import {
+  getSharedRemoteMediaWindowCache,
+  isRemoteMediaWindowSourcePath,
+  type RemoteMediaWindowCache,
+  type RemoteMediaWindowRange,
+} from './core/services/remote-media-window-cache';
 
 const log = createLogger('media');
 const AUDIO_NORMALIZATION_FILTER = 'loudnorm=I=-23:TP=-2:LRA=11';
@@ -86,6 +92,11 @@ export interface MediaGeneratorOptions {
   logDebug?: (message: string) => void;
   now?: () => number;
   execFile?: MediaGeneratorExecFile;
+  /**
+   * Local window cache for http(s) sources. Defaults to the process-wide cache shared
+   * with the timing review; pass `null` to always read remote sources directly.
+   */
+  remoteMediaWindows?: RemoteMediaWindowCache | null;
 }
 
 function sanitizeDebugToken(value: string, fallback: string): string {
@@ -232,6 +243,54 @@ export class MediaGenerator {
     }, delayMs);
   }
 
+  /**
+   * Swaps an http(s) input for the locally cached window that covers `range`, so the
+   * clip is downloaded once instead of per FFmpeg run. `acquire` downloads on a miss;
+   * `lookup` only reuses a window that another step already fetched. Any failure falls
+   * back to reading the remote source directly.
+   */
+  private async resolveRemoteWindowInput(
+    input: MediaInput,
+    range: RemoteMediaWindowRange,
+    audioStreamIndex: number | null | undefined,
+    mode: 'acquire' | 'lookup',
+  ): Promise<MediaInput> {
+    const cache =
+      this.options.remoteMediaWindows === undefined
+        ? getSharedRemoteMediaWindowCache()
+        : this.options.remoteMediaWindows;
+    const sourcePath = typeof input === 'string' ? input : input.path;
+    if (!cache || !isRemoteMediaWindowSourcePath(sourcePath)) {
+      return input;
+    }
+    const source = {
+      path: sourcePath,
+      ...(typeof input === 'object' && input.inputOptions
+        ? { inputOptions: input.inputOptions }
+        : {}),
+      audioStreamIndex:
+        typeof input === 'object' && input.singleResolvedStream ? null : (audioStreamIndex ?? null),
+    };
+    const description = describeMediaInputForDebugLog(input);
+    try {
+      const window =
+        mode === 'acquire' ? await cache.acquire(source, range) : await cache.lookup(source, range);
+      if (!window) {
+        this.logMediaDebug(`window miss ${description} mode=${mode}`);
+        return input;
+      }
+      this.logMediaDebug(
+        `window hit ${description} mode=${mode} start=${window.startTime} end=${window.endTime}`,
+      );
+      return window.media;
+    } catch (error) {
+      this.logMediaDebug(
+        `window failed ${description} mode=${mode} reason=${sanitizeDebugToken((error as Error).message, 'error')}`,
+      );
+      return input;
+    }
+  }
+
   private ffmpegError(label: string, error: ExecFileException): Error {
     if (error.code === 'ENOENT') {
       return new Error('FFmpeg not found. Install FFmpeg to enable media generation.');
@@ -281,7 +340,13 @@ export class MediaGenerator {
     const safePadding = Number.isFinite(padding) ? Math.max(0, padding) : 0;
     const start = Math.max(0, startTime - safePadding);
     const duration = endTime - start + safePadding;
-    const mediaInput = normalizeMediaInput(videoPath);
+    const sourceInput = await this.resolveRemoteWindowInput(
+      videoPath,
+      { startTime: start, endTime: start + duration },
+      audioStreamIndex,
+      'acquire',
+    );
+    const mediaInput = normalizeMediaInput(sourceInput);
     const inputDescription = describeMediaInputForDebugLog(videoPath);
     const hasSelectedAudioStream =
       !mediaInput.singleResolvedStream &&
@@ -385,7 +450,15 @@ export class MediaGenerator {
       png: 'png',
       webp: 'webp',
     };
-    const mediaInput = normalizeMediaInput(videoPath);
+    // A single frame is cheap to fetch remotely, so only reuse a window another step downloaded.
+    const mediaInput = normalizeMediaInput(
+      await this.resolveRemoteWindowInput(
+        videoPath,
+        { startTime: timestamp, endTime: timestamp },
+        null,
+        'lookup',
+      ),
+    );
     const inputDescription = describeMediaInputForDebugLog(videoPath);
 
     const args: string[] = [
@@ -533,9 +606,17 @@ export class MediaGenerator {
       );
     }
 
+    const mediaInput = normalizeMediaInput(
+      await this.resolveRemoteWindowInput(
+        videoPath,
+        { startTime: start, endTime: start + duration },
+        null,
+        'acquire',
+      ),
+    );
+
     return new Promise((resolve, reject) => {
       const outputPath = this.createTempOutputPath('animation', 'avif');
-      const mediaInput = normalizeMediaInput(videoPath);
       const startedAt = this.nowMs();
 
       const encoderArgs: string[] = ['-c:v', av1Encoder];

src/media-input.ts

@@ -11,6 +11,12 @@ export type MediaInput =
       source?: string;
       inputOptions?: MediaInputOptions;
       singleResolvedStream?: boolean;
+      /**
+       * The file keeps the original media timestamps instead of starting at zero (a
+       * stream-copied window of a longer source). Seek with `-ss` against those
+       * absolute timestamps rather than relative to the file's own start time.
+       */
+      absoluteTimestamps?: boolean;
     };
 
 export type NormalizedMediaInput = {
@@ -89,6 +95,10 @@ export function normalizeMediaInput(input: MediaInput): NormalizedMediaInput {
     inputArgs.push('-headers', headers);
   }
 
+  if (input.absoluteTimestamps) {
+    inputArgs.push('-seek_timestamp', '1');
+  }
+
   return {
     path: input.path,
     inputArgs,

src/preload.ts

@@ -191,6 +191,10 @@ const onOpenMediaTimingReviewEvent =
     IPC_CHANNELS.event.mediaTimingReviewOpen,
     (payload) => payload as MediaTimingReviewOpenPayload,
   );
+const onMediaTimingReviewPreviewEndedEvent = createQueuedIpcListenerWithPayload<string>(
+  IPC_CHANNELS.event.mediaTimingReviewPreviewEnded,
+  (payload) => (typeof payload === 'string' ? payload : ''),
+);
 const onOpenPlaylistBrowserEvent = createQueuedIpcListener(IPC_CHANNELS.event.playlistBrowserOpen);
 const onCancelYoutubeTrackPickerEvent = createQueuedIpcListener(
   IPC_CHANNELS.event.youtubePickerCancel,
@@ -469,6 +473,7 @@ const electronAPI: ElectronAPI = {
   onOpenTsukihime: onOpenTsukihimeEvent,
   onOpenYoutubeTrackPicker: onOpenYoutubeTrackPickerEvent,
   onOpenMediaTimingReview: onOpenMediaTimingReviewEvent,
+  onMediaTimingReviewPreviewEnded: onMediaTimingReviewPreviewEndedEvent,
   previewMediaTimingReview: (
     request: MediaTimingReviewPreviewRequest,
   ): Promise<MediaTimingReviewActionResult> =>

src/renderer/index.html

@@ -499,8 +499,8 @@
             <div id="kikuSelectionStep">
               <div class="kiku-info-text">
                 A card with the same expression already exists. Select which card to keep. The other
-                card's content will be merged using Kiku field grouping. You can choose whether to
-                delete the duplicate.
+                card's content will be merged using field grouping. You can choose whether to delete
+                the duplicate.
               </div>
               <div class="kiku-cards-container">
                 <div id="kikuCard1" class="kiku-card active" tabindex="0">

src/renderer/modals/media-timing-review.test.ts

@@ -8,6 +8,7 @@ import {
   formatMediaTimingTimestamp,
   mediaTimingTimeFromPointer,
   slideMediaTimingSelection,
+  trimMediaTimingSelectionEnd,
 } from './media-timing-review';
 
 test('waveform path mirrors normalized peaks around its center line', () => {
@@ -122,3 +123,35 @@ test('preview request guard blocks overlap and invalidates stale responses', ()
   guard.finish(second!);
   assert.equal(guard.isInFlight(), false);
 });
+
+test('trailing silence trim follows the last speech slice and keeps cut-off or silent lines', () => {
+  // 10 slices over a 10 s timeline: one slice per second, speech in seconds 2-4 only.
+  const peaks = [0, 0, 0.9, 0.8, 0.7, 0.1, 0.2, 0, 0, 0];
+  const base = { peaks, timelineStart: 0, timelineEnd: 10, lineStart: 2, lineEnd: 8 };
+
+  const trimmed = trimMediaTimingSelectionEnd({ ...base, selectionEnd: 8, endPadSeconds: 0 });
+  assert.ok(trimmed !== null && Math.abs(trimmed - 5.15) < 1e-9);
+
+  const padded = trimMediaTimingSelectionEnd({ ...base, selectionEnd: 8.5, endPadSeconds: 0.5 });
+  assert.ok(padded !== null && Math.abs(padded - 5.65) < 1e-9);
+
+  // Speech running through the line end means the subtitle cuts the audio off; keep it.
+  assert.equal(
+    trimMediaTimingSelectionEnd({ ...base, lineEnd: 5, selectionEnd: 5, endPadSeconds: 0 }),
+    null,
+  );
+  // No speech inside the line at all: keep the subtitle timing rather than guess.
+  assert.equal(
+    trimMediaTimingSelectionEnd({ ...base, lineStart: 6, selectionEnd: 8, endPadSeconds: 0 }),
+    null,
+  );
+  // A saving below the minimum trim is not worth moving the handle for.
+  assert.equal(
+    trimMediaTimingSelectionEnd({ ...base, lineEnd: 5.2, selectionEnd: 5.2, endPadSeconds: 0 }),
+    null,
+  );
+  assert.equal(
+    trimMediaTimingSelectionEnd({ ...base, peaks: [1], selectionEnd: 8, endPadSeconds: 0 }),
+    null,
+  );
+});

src/renderer/modals/media-timing-review.ts

@@ -11,6 +11,16 @@ const FINE_ADJUST_SECONDS = 0.1;
 const COARSE_ADJUST_SECONDS = 0.5;
 const TIMELINE_EXPANSION_SECONDS = 2;
 const LINE_REVEAL_MARGIN_SECONDS = 1;
+/**
+ * Subtitles usually linger past the dialogue for readability, so an untouched clip end
+ * follows the last speech-weighted waveform slice above this level (0..1, relative to the
+ * clip's own noise floor) plus a short tail, when that saves at least the minimum trim.
+ */
+const SPEECH_LEVEL_THRESHOLD = 0.3;
+const SPEECH_TAIL_SECONDS = 0.15;
+const MINIMUM_TRAILING_TRIM_SECONDS = 0.1;
+/** Slack past the clip length before the UI gives up waiting for mpv's end-of-clip signal. */
+const PREVIEW_END_GRACE_MS = 2_500;
 
 function clamp(value: number, minimum: number, maximum: number): number {
   return Math.min(maximum, Math.max(minimum, value));
@@ -105,6 +115,48 @@ export function buildMediaTimingWaveformPath(peaks: number[]): string {
   return `M ${upper.join(' L ')} L ${lower.join(' L ')} Z`;
 }
 
+/**
+ * Where an untouched clip should end once the waveform is known: just after the line's
+ * last speech slice, plus the configured end padding. Null keeps the subtitle timing when
+ * no speech shows inside the line, when speech runs through the line end (the subtitle is
+ * cutting the audio off, not lingering), or when the saving is too small to matter.
+ */
+export function trimMediaTimingSelectionEnd(options: {
+  peaks: readonly number[];
+  timelineStart: number;
+  timelineEnd: number;
+  lineStart: number;
+  lineEnd: number;
+  selectionEnd: number;
+  endPadSeconds: number;
+}): number | null {
+  const pointCount = options.peaks.length;
+  const span = options.timelineEnd - options.timelineStart;
+  if (pointCount < 2 || span <= 0 || options.lineEnd <= options.lineStart) return null;
+  // Point i covers [i, i + 1) / pointCount of the timeline (see computeWaveformPeaks).
+  const sliceEnd = (index: number): number =>
+    options.timelineStart + ((index + 1) / pointCount) * span;
+  const firstIndex = Math.max(
+    0,
+    Math.floor(((options.lineStart - options.timelineStart) / span) * pointCount),
+  );
+  const lastIndex = Math.min(
+    pointCount - 1,
+    Math.ceil(((options.lineEnd - options.timelineStart) / span) * pointCount) - 1,
+  );
+  let lastSpeechIndex = -1;
+  for (let index = lastIndex; index >= firstIndex; index -= 1) {
+    if ((options.peaks[index] ?? 0) >= SPEECH_LEVEL_THRESHOLD) {
+      lastSpeechIndex = index;
+      break;
+    }
+  }
+  if (lastSpeechIndex === -1 || lastSpeechIndex >= lastIndex) return null;
+  const trimmedEnd = sliceEnd(lastSpeechIndex) + SPEECH_TAIL_SECONDS + options.endPadSeconds;
+  if (options.selectionEnd - trimmedEnd < MINIMUM_TRAILING_TRIM_SECONDS) return null;
+  return trimmedEnd;
+}
+
 export function constrainMediaTimingSelection(options: {
   nextStart: number;
   nextEnd: number;
@@ -174,6 +226,8 @@ export function createMediaTimingReviewModal(
   let nextCount = 0;
   let startPadSeconds = 0;
   let endPadSeconds = 0;
+  /** True until the user moves the clip; the first waveform then trims trailing silence. */
+  let trailingTrimPending = false;
   let resolveInFlight = false;
   let previewPlaying = false;
   let previewTimer: ReturnType<typeof setTimeout> | null = null;
@@ -209,7 +263,11 @@ export function createMediaTimingReviewModal(
     previewTimer = null;
   }
 
-  /** Drives the play button label plus the playhead sweep that mirrors the hidden audio player. */
+  /**
+   * Drives the play button label plus the playhead sweep that mirrors the hidden audio player.
+   * mpv reports when the clip actually finishes (see handlePreviewEnded), which accounts for
+   * output latency such as Bluetooth headphones; the timer only covers a player that never does.
+   */
   function setPreviewPlaying(playing: boolean): void {
     previewPlaying = playing;
     ctx.dom.mediaTimingReviewPlayLabel.textContent = playing ? 'Stop preview' : 'Play selection';
@@ -222,7 +280,14 @@ export function createMediaTimingReviewModal(
     track.style.setProperty('--playhead-duration', `${clipSeconds}s`);
     void track.offsetWidth;
     track.classList.add('is-previewing');
-    previewTimer = setTimeout(() => stopPreview(), clipSeconds * 1000);
+    previewTimer = setTimeout(() => stopPreview(), clipSeconds * 1000 + PREVIEW_END_GRACE_MS);
+  }
+
+  /** The hidden player reached the end of the clip and paused itself. */
+  function handlePreviewEnded(reviewId: string): void {
+    if (!payload || payload.reviewId !== reviewId || !previewPlaying) return;
+    setPreviewPlaying(false);
+    setStatus('');
   }
 
   /** Callers that need to report a failure set their own status after stopping the preview. */
@@ -322,6 +387,10 @@ export function createMediaTimingReviewModal(
       ) {
         return;
       }
+      if (!result.ok && result.stale) {
+        closeResolvedReview();
+        return;
+      }
       const path = result.ok ? buildMediaTimingWaveformPath(result.peaks ?? []) : '';
       if (!path) {
         setWaveformState('unavailable');
@@ -329,6 +398,7 @@ export function createMediaTimingReviewModal(
       }
       ctx.dom.mediaTimingReviewWaveformPath.setAttribute('d', path);
       setWaveformState('ready');
+      trimTrailingSilence(result.peaks ?? []);
     } catch {
       if (sequence === waveformSequence && payload?.reviewId === reviewId) {
         setWaveformState('unavailable');
@@ -336,6 +406,29 @@ export function createMediaTimingReviewModal(
     }
   }
 
+  /**
+   * Moves an untouched clip end back to where the line's dialogue ends. The Line end
+   * rail keeps marking the subtitle timing, and Reset restores it.
+   */
+  function trimTrailingSilence(peaks: readonly number[]): void {
+    if (!payload || !trailingTrimPending || previewPlaying || previewRequest.isInFlight()) {
+      return;
+    }
+    const lineRange = currentLineSelection();
+    const trimmedEnd = trimMediaTimingSelectionEnd({
+      peaks,
+      timelineStart,
+      timelineEnd,
+      lineStart: lineRange.rangeStart,
+      lineEnd: lineRange.rangeEnd,
+      selectionEnd,
+      endPadSeconds,
+    });
+    if (trimmedEnd === null) return;
+    updateSelection(selectionStart, trimmedEnd);
+    setStatus('Clip end moved to where the dialogue ends. Reset restores the subtitle timing.');
+  }
+
   function queueWaveformLoad(delayMs = 0): void {
     if (waveformTimer !== null) clearTimeout(waveformTimer);
     waveformSequence += 1;
@@ -443,6 +536,7 @@ export function createMediaTimingReviewModal(
     });
     selectionStart = nextSelection.start;
     selectionEnd = nextSelection.end;
+    trailingTrimPending = false;
     if (previewPlaying || previewRequest.isInFlight()) stopPreview();
     setStatus('');
     renderSelection();
@@ -617,11 +711,13 @@ export function createMediaTimingReviewModal(
         reviewId: payload.reviewId,
         decision,
       });
-      if (!result.ok) {
+      if (!result.ok && !result.stale) {
         setStatus(result.message ?? 'The timing review could not be resolved.', true);
         showEditor();
         return;
       }
+      // A stale review was already settled by main; keeping the modal up would leave
+      // controls that can never succeed over a live mpv window.
       closeResolvedReview();
     } catch (error) {
       setStatus(error instanceof Error ? error.message : String(error), true);
@@ -669,6 +765,10 @@ export function createMediaTimingReviewModal(
         }
         return;
       }
+      if (!result.ok && result.stale) {
+        closeResolvedReview();
+        return;
+      }
       if (!result.ok) {
         setStatus(result.message ?? 'Audio preview is unavailable.', true);
         setPreviewPlaying(false);
@@ -704,6 +804,7 @@ export function createMediaTimingReviewModal(
     nextCount = 0;
     startPadSeconds = Math.max(0, nextPayload.originalStartTime - nextPayload.selectionStartTime);
     endPadSeconds = Math.max(0, nextPayload.selectionEndTime - nextPayload.originalEndTime);
+    trailingTrimPending = true;
     resolveInFlight = false;
     setPreviewPlaying(false);
     ctx.dom.mediaTimingReviewKind.textContent =
@@ -851,6 +952,7 @@ export function createMediaTimingReviewModal(
 
   return {
     openMediaTimingReviewModal,
+    handlePreviewEnded,
     requestCancel,
     handleMediaTimingReviewKeydown,
     wireDomEvents,

src/renderer/modals/tsukihime.test.ts

@@ -40,13 +40,19 @@ function createElementStub() {
 }
 
 function createListStub() {
-  return {
-    innerHTML: '',
+  const list = {
     children: [] as unknown[],
     appendChild(child: unknown) {
-      this.children.push(child);
+      list.children.push(child);
     },
   };
+  // The modal clears lists through innerHTML before re-rendering.
+  return Object.defineProperty(list, 'innerHTML', {
+    get: () => '',
+    set: () => {
+      list.children.length = 0;
+    },
+  }) as typeof list & { innerHTML: string };
 }
 
 function createTabStub(active: boolean) {
@@ -118,6 +124,7 @@ function createModalHarness(
   files: TsukihimeSubtitleFile[],
   options: {
     secondaryLanguages?: string[];
+    secondaryLanguagesGate?: Promise<void>;
     downloadFile?: (query: unknown) => Promise<unknown>;
     listFiles?: (entryId: number) => Promise<unknown>;
     searchEntries?: (query: unknown) => Promise<unknown>;
@@ -138,7 +145,10 @@ function createModalHarness(
       if (options.downloadFile) return options.downloadFile(query);
       return { ok: true, path: '/tmp/subtitles/episode01.en.ass' };
     },
-    tsukihimeGetSecondaryLanguages: async () => options.secondaryLanguages ?? ['en', 'eng'],
+    tsukihimeGetSecondaryLanguages: async () => {
+      await options.secondaryLanguagesGate;
+      return options.secondaryLanguages ?? ['en', 'eng'];
+    },
     tsukihimeListFiles: async ({ entryId }: { entryId: number }) =>
       options.listFiles ? options.listFiles(entryId) : { ok: true, data: [] },
     tsukihimeSearchEntries: async (query: unknown) =>
@@ -613,3 +623,202 @@ test('renderFiles omits the size detail when the API does not report one', () =>
     harness.restoreGlobals();
   }
 });
+
+const ENGLISH_ONLY_ENTRY = {
+  id: 606713,
+  title: 'english only release',
+  timestamp: null,
+  totalSize: null,
+  numFiles: 1,
+  sublangs: ['en'],
+};
+
+const MULTI_SUB_ENTRY = {
+  id: 12255,
+  title: 'multi-sub release',
+  timestamp: null,
+  totalSize: null,
+  numFiles: 1,
+  sublangs: ['en-US', 'ja'],
+};
+
+const UNLABELED_ENTRY = {
+  id: 12256,
+  title: 'release without langs',
+  timestamp: null,
+  totalSize: null,
+  numFiles: 1,
+  sublangs: [],
+};
+
+function visibleEntryTitles(harness: ModalHarness): string[] {
+  return (harness.entriesList.children as Array<{ textContent: string }>).map(
+    (li) => li.textContent,
+  );
+}
+
+test('Japanese tab lists only releases that carry Japanese subtitles', async () => {
+  const SECOND_JAPANESE_TRACK: TsukihimeSubtitleFile = {
+    ...JAPANESE_TRACK,
+    attachmentId: 1955401,
+    filename: 'episode01.jpn.sdh.ass',
+  };
+  const harness = createModalHarness([], {
+    // Two tracks so the modal does not auto-download a lone match.
+    listFiles: async () => ({ ok: true, data: [JAPANESE_TRACK, SECOND_JAPANESE_TRACK] }),
+  });
+  try {
+    harness.state.currentTsukihimeEntryId = null;
+    harness.state.tsukihimeEntries = [ENGLISH_ONLY_ENTRY, MULTI_SUB_ENTRY, UNLABELED_ENTRY];
+
+    pressKey(harness, 'ArrowRight');
+    assert.deepEqual(visibleEntryTitles(harness), ['multi-sub release']);
+
+    // Enter addresses the visible list, so it must pick the multi-sub release
+    // rather than the hidden first search result.
+    pressKey(harness, 'Enter');
+    await flushAsyncWork();
+    assert.equal(harness.state.currentTsukihimeEntryId, MULTI_SUB_ENTRY.id);
+    assert.equal(harness.status.textContent, 'Select a subtitle track.');
+
+    pressKey(harness, 'ArrowLeft');
+    assert.deepEqual(visibleEntryTitles(harness), [
+      'english only release',
+      'multi-sub release',
+      'release without langs',
+    ]);
+    assert.equal(harness.state.currentTsukihimeEntryId, MULTI_SUB_ENTRY.id);
+    assert.equal(harness.state.selectedTsukihimeEntryIndex, 1);
+  } finally {
+    harness.restoreGlobals();
+  }
+});
+
+test('Japanese tab reports when no release carries Japanese subtitles', () => {
+  const harness = createModalHarness([]);
+  try {
+    harness.state.currentTsukihimeEntryId = null;
+    harness.state.tsukihimeEntries = [ENGLISH_ONLY_ENTRY, UNLABELED_ENTRY];
+
+    pressKey(harness, 'ArrowRight');
+    assert.deepEqual(visibleEntryTitles(harness), []);
+    assert.equal(
+      harness.status.textContent,
+      'No releases with Japanese subtitles. Switch to the English tab.',
+    );
+
+    pressKey(harness, 'ArrowLeft');
+    assert.deepEqual(visibleEntryTitles(harness), [
+      'english only release',
+      'release without langs',
+    ]);
+    assert.equal(harness.status.textContent, 'Select a release.');
+  } finally {
+    harness.restoreGlobals();
+  }
+});
+
+test('search reports when no release carries the secondary language', async () => {
+  const harness = createModalHarness([], {
+    searchEntries: async () => ({
+      ok: true,
+      data: [{ ...MULTI_SUB_ENTRY, sublangs: ['ja'] }],
+    }),
+  });
+  try {
+    harness.state.currentTsukihimeEntryId = null;
+    harness.titleInput.value = 'Futsutsuka na Akujo';
+
+    pressKey(harness, 'Enter');
+    await flushAsyncWork();
+    assert.deepEqual(visibleEntryTitles(harness), []);
+    assert.equal(
+      harness.status.textContent,
+      'No releases with English subtitles. Switch to the Japanese tab.',
+    );
+
+    pressKey(harness, 'ArrowRight');
+    assert.deepEqual(visibleEntryTitles(harness), ['multi-sub release']);
+  } finally {
+    harness.restoreGlobals();
+  }
+});
+
+test('switching to a tab that hides the selected release clears its tracks', () => {
+  const harness = createModalHarness([ENGLISH_TRACK, JAPANESE_TRACK]);
+  try {
+    harness.state.tsukihimeEntries = [ENGLISH_ONLY_ENTRY, MULTI_SUB_ENTRY];
+
+    pressKey(harness, 'ArrowRight');
+    assert.equal(harness.state.currentTsukihimeEntryId, null);
+    assert.deepEqual(harness.state.tsukihimeFiles, []);
+    assert.deepEqual(visibleEntryTitles(harness), ['multi-sub release']);
+    assert.equal(harness.status.textContent, 'Select a release.');
+  } finally {
+    harness.restoreGlobals();
+  }
+});
+
+test('a search waits for the configured secondary languages before filtering', async () => {
+  let openGate!: () => void;
+  const harness = createModalHarness([], {
+    secondaryLanguages: ['de'],
+    secondaryLanguagesGate: new Promise<void>((resolve) => {
+      openGate = resolve;
+    }),
+    searchEntries: async () => ({
+      ok: true,
+      data: [{ ...MULTI_SUB_ENTRY, title: 'german release', sublangs: ['de'] }],
+    }),
+  });
+  try {
+    harness.state.tsukihimeModalOpen = false;
+    harness.modal.openTsukihimeModal();
+    harness.titleInput.value = 'Futsutsuka na Akujo';
+
+    // Searching before the config arrives must not filter against the English
+    // fallback, which would hide this German-only release.
+    pressKey(harness, 'Enter');
+    await flushAsyncWork();
+    assert.deepEqual(visibleEntryTitles(harness), []);
+
+    openGate();
+    await flushAsyncWork();
+
+    assert.deepEqual(visibleEntryTitles(harness), ['german release']);
+  } finally {
+    harness.restoreGlobals();
+  }
+});
+
+test('a search from a prior modal session cannot repopulate a reopened modal', async () => {
+  let openGate!: () => void;
+  const harness = createModalHarness([], {
+    secondaryLanguagesGate: new Promise<void>((resolve) => {
+      openGate = resolve;
+    }),
+    searchEntries: async () => ({ ok: true, data: [MULTI_SUB_ENTRY] }),
+  });
+  try {
+    harness.state.tsukihimeModalOpen = false;
+    harness.modal.openTsukihimeModal();
+    harness.titleInput.value = 'Futsutsuka na Akujo';
+
+    // The search parks on the language config, then the user closes and
+    // reopens the modal before it resolves.
+    pressKey(harness, 'Enter');
+    harness.modal.closeTsukihimeModal();
+    harness.modal.openTsukihimeModal();
+    await flushAsyncWork();
+    harness.status.textContent = 'Fresh modal session';
+
+    openGate();
+    await flushAsyncWork();
+
+    assert.deepEqual(harness.state.tsukihimeEntries, []);
+    assert.deepEqual(visibleEntryTitles(harness), []);
+    assert.equal(harness.status.textContent, 'Fresh modal session');
+  } finally {
+    harness.restoreGlobals();
+  }
+});

src/renderer/modals/tsukihime.ts

@@ -41,7 +41,13 @@ export function createTsukihimeModal(
 
   // Defaults to English until the configured secondary languages arrive.
   let secondaryLanguages: string[] = ['en'];
+  // Both tab filters read the configured languages, so a search must wait for
+  // them rather than filtering against the English fallback.
+  let secondaryLanguagesReady: Promise<void> = Promise.resolve();
   let activeDownloadToken = 0;
+  // Bumped by every new search and by closing the modal, so results that
+  // arrive late cannot repopulate a reopened modal or a newer search.
+  let activeSearchToken = 0;
 
   function secondaryTabLabel(): string {
     return describeTsukihimeTabLanguages(secondaryLanguages);
@@ -61,6 +67,47 @@ export function createTsukihimeModal(
     );
   }
 
+  // Releases are filtered by the languages the search index reports for
+  // them. Most releases carry no Japanese track, so the primary tab hides
+  // them outright. A release with no language data cannot be classified and
+  // stays on the secondary tab, mirroring how unlabeled tracks are handled.
+  function entryMatchesTab(entry: TsukihimeEntry, tab: 'secondary' | 'primary'): boolean {
+    if (tab === 'primary') {
+      return entry.sublangs.some((lang) => normalizeTsukihimeLangCode(lang) === 'ja');
+    }
+    if (entry.sublangs.length === 0) return true;
+    return entry.sublangs.some(
+      (lang) =>
+        normalizeTsukihimeLangCode(lang) !== 'ja' &&
+        tsukihimeTrackMatchesLanguages(lang, secondaryLanguages),
+    );
+  }
+
+  function getVisibleEntries(): TsukihimeEntry[] {
+    return ctx.state.tsukihimeEntries.filter((entry) =>
+      entryMatchesTab(entry, ctx.state.tsukihimeActiveTab),
+    );
+  }
+
+  function describeEmptyReleases(): string {
+    const otherTab = ctx.state.tsukihimeActiveTab === 'primary' ? 'secondary' : 'primary';
+    const otherTabHasReleases = ctx.state.tsukihimeEntries.some((entry) =>
+      entryMatchesTab(entry, otherTab),
+    );
+    const language = ctx.state.tsukihimeActiveTab === 'primary' ? 'Japanese' : secondaryTabLabel();
+    const otherLabel = otherTab === 'primary' ? 'Japanese' : secondaryTabLabel();
+    return otherTabHasReleases
+      ? `No releases with ${language} subtitles. Switch to the ${otherLabel} tab.`
+      : `No releases with ${language} subtitles.`;
+  }
+
+  function clearFiles(): void {
+    ctx.state.tsukihimeFiles = [];
+    ctx.state.selectedTsukihimeFileIndex = 0;
+    ctx.dom.tsukihimeFilesList.innerHTML = '';
+    ctx.dom.tsukihimeFilesSection.classList.add('hidden');
+  }
+
   function renderTabs(): void {
     const primaryActive = ctx.state.tsukihimeActiveTab === 'primary';
     ctx.dom.tsukihimeTabSecondaryButton.setAttribute(
@@ -98,12 +145,37 @@ export function createTsukihimeModal(
     ctx.state.selectedTsukihimeFileIndex = 0;
     renderTabs();
 
-    if (ctx.state.tsukihimeFiles.length === 0) return;
-    renderFiles();
-    if (getVisibleFiles().length === 0) {
-      setTsukihimeStatus(describeEmptyTab());
-    } else {
-      setTsukihimeStatus('Select a subtitle track.');
+    const currentEntry = ctx.state.tsukihimeEntries.find(
+      (entry) => entry.id === ctx.state.currentTsukihimeEntryId,
+    );
+    if (currentEntry && !entryMatchesTab(currentEntry, tab)) {
+      // The selected release is hidden on this tab; drop its tracks so the
+      // list matches what the tab claims to show.
+      ctx.state.currentTsukihimeEntryId = null;
+      ctx.state.selectedTsukihimeEntryIndex = 0;
+      clearFiles();
+      renderEntries();
+      setTsukihimeStatus(
+        getVisibleEntries().length === 0 ? describeEmptyReleases() : 'Select a release.',
+      );
+      return;
+    }
+
+    const visibleEntries = getVisibleEntries();
+    ctx.state.selectedTsukihimeEntryIndex = currentEntry ? visibleEntries.indexOf(currentEntry) : 0;
+    renderEntries();
+
+    if (ctx.state.tsukihimeFiles.length > 0) {
+      renderFiles();
+      setTsukihimeStatus(
+        getVisibleFiles().length === 0 ? describeEmptyTab() : 'Select a subtitle track.',
+      );
+      return;
+    }
+    if (!currentEntry && ctx.state.tsukihimeEntries.length > 0) {
+      setTsukihimeStatus(
+        visibleEntries.length === 0 ? describeEmptyReleases() : 'Select a release.',
+      );
     }
   }
 
@@ -121,13 +193,14 @@ export function createTsukihimeModal(
 
   function renderEntries(): void {
     ctx.dom.tsukihimeEntriesList.innerHTML = '';
-    if (ctx.state.tsukihimeEntries.length === 0) {
+    const visibleEntries = getVisibleEntries();
+    if (visibleEntries.length === 0) {
       ctx.dom.tsukihimeEntriesSection.classList.add('hidden');
       return;
     }
 
     ctx.dom.tsukihimeEntriesSection.classList.remove('hidden');
-    ctx.state.tsukihimeEntries.forEach((entry, index) => {
+    visibleEntries.forEach((entry, index) => {
       const li = document.createElement('li');
       li.textContent = entry.title;
 
@@ -210,11 +283,15 @@ export function createTsukihimeModal(
       return;
     }
 
+    const searchToken = ++activeSearchToken;
     resetTsukihimeLists();
     setTsukihimeStatus('Searching TsukiHime...');
+    await secondaryLanguagesReady;
+    if (searchToken !== activeSearchToken) return;
 
     const response: TsukihimeApiResponse<TsukihimeEntry[]> =
       await window.electronAPI.tsukihimeSearchEntries({ query });
+    if (searchToken !== activeSearchToken) return;
     if (!response.ok) {
       setTsukihimeStatus(response.error.error, true);
       return;
@@ -228,20 +305,22 @@ export function createTsukihimeModal(
       return;
     }
 
+    const visibleEntries = getVisibleEntries();
+    if (visibleEntries.length === 0) {
+      setTsukihimeStatus(describeEmptyReleases());
+      return;
+    }
+
     setTsukihimeStatus('Select a release.');
     renderEntries();
-    if (ctx.state.tsukihimeEntries.length === 1) {
+    if (visibleEntries.length === 1) {
       selectEntry(0);
     }
   }
 
   async function loadFiles(entryId: number): Promise<void> {
     setTsukihimeStatus('Loading subtitle tracks...');
-    ctx.state.tsukihimeFiles = [];
-    ctx.state.selectedTsukihimeFileIndex = 0;
-
-    ctx.dom.tsukihimeFilesList.innerHTML = '';
-    ctx.dom.tsukihimeFilesSection.classList.add('hidden');
+    clearFiles();
 
     const response: TsukihimeApiResponse<TsukihimeSubtitleFile[]> =
       await window.electronAPI.tsukihimeListFiles({ entryId });
@@ -279,11 +358,14 @@ export function createTsukihimeModal(
     }
   }
 
+  // `index` addresses the entries visible on the active tab, not the full
+  // search result list.
   function selectEntry(index: number): void {
-    if (index < 0 || index >= ctx.state.tsukihimeEntries.length) return;
+    const visibleEntries = getVisibleEntries();
+    if (index < 0 || index >= visibleEntries.length) return;
 
     ctx.state.selectedTsukihimeEntryIndex = index;
-    ctx.state.currentTsukihimeEntryId = ctx.state.tsukihimeEntries[index]!.id;
+    ctx.state.currentTsukihimeEntryId = visibleEntries[index]!.id;
     renderEntries();
 
     if (ctx.state.currentTsukihimeEntryId !== null) {
@@ -363,16 +445,15 @@ export function createTsukihimeModal(
     resetTsukihimeLists();
     renderTabs();
 
-    const secondaryLanguagesReady = loadSecondaryLanguages();
+    secondaryLanguagesReady = loadSecondaryLanguages();
 
     window.electronAPI
       .getJimakuMediaInfo()
-      .then(async (info: JimakuMediaInfo) => {
+      .then((info: JimakuMediaInfo) => {
         ctx.dom.tsukihimeTitleInput.value = info.title || '';
         ctx.dom.tsukihimeEpisodeInput.value = info.episode ? String(info.episode) : '';
 
         if (info.confidence === 'high' && info.title && info.episode) {
-          await secondaryLanguagesReady;
           void performTsukihimeSearch();
         } else if (info.title) {
           setTsukihimeStatus('Check title/episode and press Search.');
@@ -389,6 +470,7 @@ export function createTsukihimeModal(
     if (!ctx.state.tsukihimeModalOpen) return;
 
     activeDownloadToken += 1;
+    activeSearchToken += 1;
     ctx.state.tsukihimeModalOpen = false;
     options.syncSettingsModalSubtitleSuppression();
     ctx.dom.tsukihimeModal.classList.add('hidden');
@@ -438,9 +520,9 @@ export function createTsukihimeModal(
           ctx.state.selectedTsukihimeFileIndex + 1,
         );
         renderFiles();
-      } else if (ctx.state.tsukihimeEntries.length > 0) {
+      } else if (getVisibleEntries().length > 0) {
         ctx.state.selectedTsukihimeEntryIndex = Math.min(
-          ctx.state.tsukihimeEntries.length - 1,
+          getVisibleEntries().length - 1,
           ctx.state.selectedTsukihimeEntryIndex + 1,
         );
         renderEntries();
@@ -456,7 +538,7 @@ export function createTsukihimeModal(
           ctx.state.selectedTsukihimeFileIndex - 1,
         );
         renderFiles();
-      } else if (ctx.state.tsukihimeEntries.length > 0) {
+      } else if (getVisibleEntries().length > 0) {
         ctx.state.selectedTsukihimeEntryIndex = Math.max(
           0,
           ctx.state.selectedTsukihimeEntryIndex - 1,
@@ -470,7 +552,7 @@ export function createTsukihimeModal(
       e.preventDefault();
       if (getVisibleFiles().length > 0) {
         void selectFile(ctx.state.selectedTsukihimeFileIndex);
-      } else if (ctx.state.tsukihimeEntries.length > 0) {
+      } else if (getVisibleEntries().length > 0) {
         selectEntry(ctx.state.selectedTsukihimeEntryIndex);
       } else {
         void performTsukihimeSearch();

src/renderer/renderer.ts

@@ -591,6 +591,11 @@ function registerModalOpenHandlers(): void {
       mediaTimingReviewModal.openMediaTimingReviewModal(payload);
     });
   });
+  window.electronAPI.onMediaTimingReviewPreviewEnded((reviewId) => {
+    runGuarded('media-timing-review:preview-ended', () => {
+      mediaTimingReviewModal.handlePreviewEnded(reviewId);
+    });
+  });
   window.electronAPI.onOpenPlaylistBrowser(() => {
     runGuardedAsync('playlist-browser:open', async () => {
       await playlistBrowserModal.openPlaylistBrowserModal();

src/shared/ipc/contracts.ts

@@ -148,6 +148,7 @@ export const IPC_CHANNELS = {
     tsukihimeOpen: 'tsukihime:open',
     youtubePickerOpen: 'youtube:picker-open',
     mediaTimingReviewOpen: 'media-timing-review:open',
+    mediaTimingReviewPreviewEnded: 'media-timing-review:preview-ended',
     youtubePickerCancel: 'youtube:picker-cancel',
     playlistBrowserOpen: 'playlist-browser:open',
     sessionNumericSelectionStart: 'session:numeric-selection-start',

src/shared/ipc/validators.ts

@@ -60,6 +60,7 @@ const RUNTIME_OPTION_IDS: RuntimeOptionId[] = [
   'subtitle.annotation.jlpt',
   'subtitle.annotation.frequency',
   'anki.kikuFieldGrouping',
+  'anki.senrenFieldGrouping',
   'anki.nPlusOneMatchMode',
 ];
 

src/subtitle-timing-tracker.ts

@@ -67,8 +67,13 @@ export class SubtitleTimingTracker {
 
     // Check for duplicate of most recent entry (deduplicate adjacent repeats)
     const lastEntry = this.history[this.history.length - 1];
-    if (lastEntry && lastEntry.timingKey === timingKey) {
-      // Update timing to most recent occurrence
+    if (
+      lastEntry &&
+      lastEntry.timingKey === timingKey &&
+      lastEntry.startTime === startTime &&
+      lastEntry.endTime === endTime
+    ) {
+      // Refresh metadata for repeated notifications of the same subtitle event.
       lastEntry.startTime = startTime;
       lastEntry.endTime = endTime;
       lastEntry.secondaryText = displaySecondaryText;
@@ -107,28 +112,20 @@ export class SubtitleTimingTracker {
   }
 
   /**
-   * Get recent subtitle blocks in chronological order.
-   * Returns the last `count` subtitle events (oldest → newest).
+   * Get recent subtitle blocks in timeline order.
+   * Returns up to `count` known subtitle events ending at the current event.
    * Blocks preserve internal line breaks and are joined with blank lines.
    */
   getRecentBlocks(count: number): string[] {
-    if (count <= 0) return [];
-    if (count > this.history.length) {
-      count = this.history.length;
-    }
-    return this.history.slice(-count).map((entry) => entry.displayText);
+    return this.getRecentTimelineEntries(count).map((entry) => entry.displayText);
   }
 
   /**
-   * Get recent subtitle blocks with their original event timings.
-   * Returns the last `count` subtitle events (oldest → newest).
+   * Get recent subtitle blocks with their original event timings in timeline order.
+   * Returns up to `count` known subtitle events ending at the current event.
    */
   getRecentEntries(count: number): SubtitleTimingBlock[] {
-    if (count <= 0) return [];
-    if (count > this.history.length) {
-      count = this.history.length;
-    }
-    return this.history.slice(-count).map((entry) => ({
+    return this.getRecentTimelineEntries(count).map((entry) => ({
       displayText: entry.displayText,
       startTime: entry.startTime,
       endTime: entry.endTime,
@@ -144,6 +141,43 @@ export class SubtitleTimingTracker {
     return lastEntry ? lastEntry.displayText : null;
   }
 
+  private getRecentTimelineEntries(count: number): HistoryEntry[] {
+    if (count <= 0) return [];
+
+    const currentEntry = this.history[this.history.length - 1];
+    if (!currentEntry) return [];
+
+    const timelineEntries: HistoryEntry[] = [];
+    for (const entry of this.history) {
+      const existingIndex = timelineEntries.findIndex((candidate) =>
+        this.isSameSubtitleEvent(candidate, entry),
+      );
+      if (existingIndex === -1) {
+        timelineEntries.push(entry);
+      } else {
+        timelineEntries[existingIndex] = entry;
+      }
+    }
+
+    timelineEntries.sort(
+      (left, right) => left.startTime - right.startTime || left.endTime - right.endTime,
+    );
+    const currentIndex = timelineEntries.findIndex((entry) =>
+      this.isSameSubtitleEvent(entry, currentEntry),
+    );
+    if (currentIndex === -1) return [];
+
+    return timelineEntries.slice(Math.max(0, currentIndex - count + 1), currentIndex + 1);
+  }
+
+  private isSameSubtitleEvent(left: HistoryEntry, right: HistoryEntry): boolean {
+    return (
+      left.timingKey === right.timingKey &&
+      left.startTime === right.startTime &&
+      left.endTime === right.endTime
+    );
+  }
+
   private findFuzzyMatch(text: string): { startTime: number; endTime: number } | null {
     let bestMatch: TimingEntry | null = null;
     let bestScore = 0;

src/types/anki.ts

@@ -79,6 +79,8 @@ export interface MediaTimingReviewResolveRequest {
 export interface MediaTimingReviewActionResult {
   ok: boolean;
   message?: string;
+  /** The review this request targeted has already ended; the renderer should close. */
+  stale?: boolean;
 }
 
 export interface NotificationOptions {
@@ -195,6 +197,11 @@ export interface AnkiConnectConfig {
     fieldGrouping?: 'auto' | 'manual' | 'disabled';
     deleteDuplicateInAuto?: boolean;
   };
+  isSenren?: {
+    enabled?: boolean;
+    fieldGrouping?: 'auto' | 'manual' | 'disabled';
+    deleteDuplicateInAuto?: boolean;
+  };
   lapisKiku?: {
     wordCardKind?: WordCardKind;
   };

src/types/config.ts

@@ -286,6 +286,11 @@ export interface ResolvedConfig {
       fieldGrouping: 'auto' | 'manual' | 'disabled';
       deleteDuplicateInAuto: boolean;
     };
+    isSenren: {
+      enabled: boolean;
+      fieldGrouping: 'auto' | 'manual' | 'disabled';
+      deleteDuplicateInAuto: boolean;
+    };
     lapisKiku: {
       wordCardKind: WordCardKind;
     };

src/types/runtime-options.ts

@@ -7,6 +7,7 @@ export type RuntimeOptionId =
   | 'subtitle.annotation.jlpt'
   | 'subtitle.annotation.frequency'
   | 'anki.kikuFieldGrouping'
+  | 'anki.senrenFieldGrouping'
   | 'anki.nPlusOneMatchMode';
 
 export type RuntimeOptionScope = 'ankiConnect' | 'subtitle';

src/types/runtime.ts

@@ -523,6 +523,7 @@ export interface ElectronAPI {
   onOpenTsukihime: (callback: () => void) => void;
   onOpenYoutubeTrackPicker: (callback: (payload: YoutubePickerOpenPayload) => void) => void;
   onOpenMediaTimingReview: (callback: (payload: MediaTimingReviewOpenPayload) => void) => void;
+  onMediaTimingReviewPreviewEnded: (callback: (reviewId: string) => void) => void;
   previewMediaTimingReview: (
     request: MediaTimingReviewPreviewRequest,
   ) => Promise<MediaTimingReviewActionResult>;