Repository navigation
docs(stt): record why transcript cuts are not snapped to quiet points - #1057
Conversation
…#1023) Measured on the word-timing harness with the CTC aligner: an energy-minimum snap of the cut point lands inside words and costs up to 16 points of clean one-word deletes; gated to near-silence it moves under 1% of boundaries and changes nothing. The gain a short gap offers, cut-breath.ts already takes.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configuration
📒 Files selected for processing (1)
Included review availability: This review used your included allowance. Your plan provides up to 8 included reviews per hour; 4 remain after this review. 📝 WalkthroughWalkthroughThe architecture documentation adds measurements from evaluating cut snapping between words with gaps under 100 ms. It compares quietest-point snapping windows with existing cut-breath behavior and records results for near-silence and RMS snapping. ChangesCut-Snapping Evaluation
Priority: ⬇️ Low Estimated code review effort: 1 (Trivial) | ~4 minutes Change: Other Suggested reviewers: Merge Risk: ⚪ Minimal · up to The documentation contains an inaccurate pointer to the RMS-snap discussion, but this is a minor issue and does not materially affect merge readiness. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Summary
Measured the quiet-point cut snap suggested in #1023 on
tools/stt-eval/word-timing(TTS corpus of 48 clips, clean and noisy, CTC aligner):cut-breath.tsalready does.The numbers now sit in
transcription-and-captions.md, with the other ideas tried and left out. The vocabulary prompt (part 1) needs the whisper-stt helper to read apromptfield and setinitial_prompt, so it is left for a follow-up with a helper rebuild.Related issue
Part of #1023
Type of change
Release impact
Desktop impact
Testing
🤖 Generated with Claude Code
Summary by CodeRabbit