Merge pull request #266 from pipecat-ai/aleix/silero-num-frames-fixes

vad: fix Silero VAD required number of frames

CHANGELOG.md

@@ -5,6 +5,19 @@ All notable changes to **pipecat** will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.0.35] - 2024-06-28
+
+### Changed
+
+- `FastAPIWebsocketParams` now require a serializer.
+
+- `TwilioFrameSerializer` now requires a `streamSid`.
+
+### Fixed
+
+- Silero VAD number of frames needs to be 512 for 16000 sample rate or 256 for
+  8000 sample rate.
+
 ## [0.0.34] - 2024-06-25
 
 ### Fixed

src/pipecat/vad/silero.py

@@ -36,6 +36,9 @@ class SileroVADAnalyzer(VADAnalyzer):
     def __init__(self, sample_rate=16000, params: VADParams = VADParams()):
         super().__init__(sample_rate=sample_rate, num_channels=1, params=params)
 
+        if sample_rate != 16000 and sample_rate != 8000:
+            raise Exception("Silero VAD sample rate needs to be 16000 or 8000")
+
         logger.debug("Loading Silero VAD model...")
 
         (self._model, utils) = torch.hub.load(
@@ -51,7 +54,7 @@ def __init__(self, sample_rate=16000, params: VADParams = VADParams()):
     #
 
     def num_frames_required(self) -> int:
-        return int(self.sample_rate / 100) * 4  # 40ms
+        return 512 if self.sample_rate == 16000 else 256
 
     def voice_confidence(self, buffer) -> float:
         try: