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

Set vad_force_turn_endpoint to False in SonioxSTTService

Browse Source
This commit is contained in:
Mark Backman
2026-02-10 09:16:03 -05:00
parent d47d95e1f0
commit 88e981c013
3 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
changelog/3697.changed.2.md Normal file
View File

@@ -0,0 +1 @@
- Update `SonioxSTTService` to set `vad_force_turn_endpoint` to `True`. This setting disabled the turn detection logic available natively in Soniox. Instead, Soniox relies on a local VAD to finalize the transcript. This configuration meaningfully reduces the time to final segment for Soniox. With this setting enabled, Soniox outputs a transcript in ~250ms (median). Pipecat enables smart-turn detection by default using the `LocalSmartTurnAnalyzerV3`. To use the native turn detection logic in Soniox, just set `vad_force_turn_endpoint` to `False`.

1
examples/foundational/07za-interruptible-soniox.py
View File

@@ -53,7 +53,6 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
stt = SonioxSTTService( stt = SonioxSTTService(
api_key=os.getenv("SONIOX_API_KEY"), api_key=os.getenv("SONIOX_API_KEY"),
vad_force_turn_endpoint=True,
params=SonioxInputParams( params=SonioxInputParams(
language_hints=[Language.EN], language_hints=[Language.EN],
language_hints_strict=True, language_hints_strict=True,

5
src/pipecat/services/soniox/stt.py
View File

@@ -152,7 +152,7 @@ class SonioxSTTService(WebsocketSTTService):
url: str = "wss://stt-rt.soniox.com/transcribe-websocket", url: str = "wss://stt-rt.soniox.com/transcribe-websocket",
sample_rate: Optional[int] = None, sample_rate: Optional[int] = None,
params: Optional[SonioxInputParams] = None, params: Optional[SonioxInputParams] = None,
vad_force_turn_endpoint: bool = False, vad_force_turn_endpoint: bool = True,
ttfs_p99_latency: Optional[float] = SONIOX_TTFS_P99, ttfs_p99_latency: Optional[float] = SONIOX_TTFS_P99,
**kwargs, **kwargs,
): ):
@@ -164,7 +164,8 @@ class SonioxSTTService(WebsocketSTTService):
sample_rate: Audio sample rate. sample_rate: Audio sample rate.
params: Additional configuration parameters, such as language hints, context and params: Additional configuration parameters, such as language hints, context and
speaker diarization. speaker diarization.
vad_force_turn_endpoint: Listen to `VADUserStoppedSpeakingFrame` to send finalize message to Soniox. If disabled, Soniox will detect the end of the speech. vad_force_turn_endpoint: Listen to `VADUserStoppedSpeakingFrame` to send finalize message to Soniox.
If disabled, Soniox will detect the end of the speech. Defaults to True.
ttfs_p99_latency: P99 latency from speech end to final transcript in seconds. ttfs_p99_latency: P99 latency from speech end to final transcript in seconds.
Override for your deployment. See https://github.com/pipecat-ai/stt-benchmark Override for your deployment. See https://github.com/pipecat-ai/stt-benchmark
**kwargs: Additional arguments passed to the STTService. **kwargs: Additional arguments passed to the STTService.