Merge pull request #1499 from pipecat-ai/aleix/base-output-transport-set-chunks-size

BaseOutputTransport: allow setting 10ms output audio chunks

CHANGELOG.md

@@ -9,6 +9,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Added `TransportParams.audio_out_10ms_chunks` parameter to allow controlling
+  the amount of audio being sent by the output transport. It defaults to 2, so
+  20ms audio chunks are sent.
+
 - Added `QwenLLMService` for Qwen integration with an OpenAI-compatible
   interface. Added foundational example `14q-function-calling-qwen.py`.
 

src/pipecat/transports/base_output.py

@@ -79,10 +79,11 @@ class BaseOutputTransport(FrameProcessor):
     async def start(self, frame: StartFrame):
         self._sample_rate = self._params.audio_out_sample_rate or frame.audio_out_sample_rate
 
-        # We will write 20ms audio at a time. If we receive long audio frames we
+        # We will write 10ms*CHUNKS of audio at a time (where CHUNKS is the
+        # `audio_out_10ms_chunks` parameter). If we receive long audio frames we
         # will chunk them. This will help with interruption handling.
         audio_bytes_10ms = int(self._sample_rate / 100) * self._params.audio_out_channels * 2
-        self._audio_chunk_size = audio_bytes_10ms * 2
+        self._audio_chunk_size = audio_bytes_10ms * self._params.audio_out_10ms_chunks
 
         # Start audio mixer.
         if self._params.audio_out_mixer:

src/pipecat/transports/base_transport.py

@@ -31,6 +31,7 @@ class TransportParams(BaseModel):
     audio_out_sample_rate: Optional[int] = None
     audio_out_channels: int = 1
     audio_out_bitrate: int = 96000
+    audio_out_10ms_chunks: int = 2
     audio_out_mixer: Optional[BaseAudioMixer] = None
     audio_in_enabled: bool = False
     audio_in_sample_rate: Optional[int] = None