Skip to content

fix: improve vad sensitivity and buffer size for better transcription accuracy - #18

Merged
MusicMaster4 merged 1 commit into
mainfrom
testing
Aug 1, 2026
Merged

fix: improve vad sensitivity and buffer size for better transcription accuracy#18
MusicMaster4 merged 1 commit into
mainfrom
testing

Conversation

@MusicMaster4

Copy link
Copy Markdown
Owner

Summary

  • Adjusted VAD sensitivity to retain more speech, improving transcription accuracy for clipped consonants and proper nouns.
  • Increased ring buffer size to 12 to preserve more audio around VAD activations.
  • Increased beam size and best_of parameters to 8 for more accurate transcription.
  • Added patience and condition_on_previous_text parameters to improve transcription quality.
  • Increased silence_frames threshold to 18 to better handle longer pauses in speech.

@MusicMaster4
MusicMaster4 merged commit 35560c5 into main Aug 1, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant