Merge pull request #3628 from pipecat-ai/mb/broadcast-speech-control-params-frame

Fix: Broadcast SpeechControlParamsFrame from VADController

src/pipecat/audio/vad/vad_controller.py

@@ -106,6 +106,8 @@ class VADController(BaseObject):
 
     async def _start(self, frame: StartFrame):
         self._vad_analyzer.set_sample_rate(frame.audio_in_sample_rate)
+        # Broadcast initial VAD params so other services (e.g. STT) can use them
+        await self.broadcast_frame(SpeechControlParamsFrame, vad_params=self._vad_analyzer.params)
 
     async def _handle_audio(self, frame: InputAudioRawFrame):
         """Process an audio chunk and emit speech events as needed.

src/pipecat/processors/audio/vad_processor.py

@@ -93,8 +93,10 @@ class VADProcessor(FrameProcessor):
         """
         await super().process_frame(frame, direction)
 
+        # Forward the frame first, then let VAD controller process. This ensures:
+        # 1. StartFrame reaches downstream before SpeechControlParamsFrame is broadcast
+        # 2. Audio flows through immediately while VAD detection happens after
+        await self.push_frame(frame, direction)
+
         # Let the VAD controller handle the frame
         await self._vad_controller.process_frame(frame)
-
-        # Always forward the frame
-        await self.push_frame(frame, direction)