wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #4337 from pipecat-ai/mb/fix-speech-stop-strategy

Browse Source 
Split user-turn stop timeout into independent speech and STT timers
This commit is contained in:
Mark Backman
2026-04-22 10:23:03 -04:00
committed by GitHub
parent 6b6896a543 dcbe86d0fc
commit 3f3d3c9203
4 changed files with 216 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
changelog/4337.changed.2.md Normal file
View File

@@ -0,0 +1 @@
- `SpeechTimeoutUserTurnStopStrategy` now waits only `user_speech_timeout` when a transcript arrives without a VAD stop event, rather than `max(ttfs_p99_latency, user_speech_timeout)`. If you had `ttfs_p99_latency > user_speech_timeout`, turn detection in that path is slightly faster than before.

1
changelog/4337.changed.md Normal file
View File

@@ -0,0 +1 @@
- If you use an STT service that emits finalized transcripts (Speechmatics, Soniox, Deepgram Flux, AssemblyAI) with `SpeechTimeoutUserTurnStopStrategy`, user turns now end as soon as `user_speech_timeout` elapses after VAD stop. Previously the strategy also waited for the STT P99 latency (`ttfs_p99_latency`) even when the transcript was already marked final. `user_speech_timeout` is still honored as a floor — STT finalization never shortens it.