sudacode/SubMiner
1
0
Fork 0
mirror of https://github.com/ksyasuda/SubMiner.git synced 2026-08-17 00:18:41 -07:00
Code Issues Packages Projects Releases Wiki Activity

feat(anki): add waveform-guided timing controls

Browse Source 
- Add speech-weighted waveform analysis and playback playhead
- Support dragging, sliding, and keyboard nudging for clip timing
This commit is contained in:
sudacode
2026-08-16 16:41:14 -07:00
parent 1b1b062803
commit e22117fe83
18 changed files with 1120 additions and 141 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs-site/anki-integration.md
+1 -1
View File
@@ -179,7 +179,7 @@ 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, provides draggable start and end handles, 100 ms adjustments, audio preview, and controls to reveal another five seconds before or after the visible timeline. 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. 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; buttons reveal another five seconds before or after the visible timeline. A speech-weighted waveform and markers for the original subtitle timing help locate sentence boundaries. 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.
Canceling the review lets you keep editing, finish with the original timing, or discard the card. Discard deletes an existing Yomitan or audio card and skips creation for a direct sentence card. Clipboard updates and stats-dashboard mining do not open timing review. Audio preview failure does not block confirmation or card creation. The option is disabled by default and hot-reloads.