Release v26.04.16.1 — Silero VAD voice detection #26

Merged
bvandeusen merged 0 commits from dev into main 2026-04-16 21:00:55 -04:00
bvandeusen commented 2026-04-16 21:00:48 -04:00 (Migrated from git.fabledsword.com)

Summary

  • Silero VAD — Replace amplitude-based silence detection with Silero VAD v5 ONNX neural network for robust speech-vs-noise classification
  • record v6 — Upgrade from record ^5.0.0 to ^6.2.0, remove record_linux override
  • No-speech guard — Show error on manual stop without detected speech instead of sending to Whisper

Test plan

  • Quiet room: speech detected, auto-stop on pause, correct transcript
  • Noisy environment: speech detected through noise, no false stops
  • Click mic, say nothing: "No speech detected" error
  • Manual stop mid-sentence: immediate stop, transcribes captured audio
  • TTS mode: VAD + TTS loop works correctly
## Summary - **Silero VAD** — Replace amplitude-based silence detection with Silero VAD v5 ONNX neural network for robust speech-vs-noise classification - **record v6** — Upgrade from record ^5.0.0 to ^6.2.0, remove record_linux override - **No-speech guard** — Show error on manual stop without detected speech instead of sending to Whisper ## Test plan - [ ] Quiet room: speech detected, auto-stop on pause, correct transcript - [ ] Noisy environment: speech detected through noise, no false stops - [ ] Click mic, say nothing: "No speech detected" error - [ ] Manual stop mid-sentence: immediate stop, transcribes captured audio - [ ] TTS mode: VAD + TTS loop works correctly
This repo is archived. You cannot comment on pull requests.
No Reviewers
No Label
1 Participants
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: bvandeusen/FabledApp#26