Merge pull request #1500 from pipecat-ai/aleix/base-output-transport-optimize-bot-speaking

BaseOutputTransport: optimize BotSpeakingFrames

CHANGELOG.md

@@ -76,17 +76,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `GladiaSTTService` now uses the `solaria-1` model by default. Other params
   use Gladia's default values. Added support for more language codes.
 
-### Fixed
-
-- Fixed an issue that could cause the `TranscriptionUpdateFrame` being pushed
-  because of an interruption to be discarded.
-
-- Fixed an issue that would cause `SegmentedSTTService` based services
-  (e.g. `OpenAISTTService`) to try to transcribe non-spoken audio, causing
-  invalid transcriptions.
-
-- Fixed an issue where `GoogleTTSService` was emitting two `TTSStoppedFrames`.
-
 ### Deprecated
 
 - All Pipecat services imports have been deprecated and a warning will be shown
@@ -104,6 +93,22 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Deprecated using `GladiaSTTService.InputParams` directly. Use the new
   `GladiaInputParams` class instead.
 
+### Fixed
+
+- Fixed an issue that could cause the `TranscriptionUpdateFrame` being pushed
+  because of an interruption to be discarded.
+
+- Fixed an issue that would cause `SegmentedSTTService` based services
+  (e.g. `OpenAISTTService`) to try to transcribe non-spoken audio, causing
+  invalid transcriptions.
+
+- Fixed an issue where `GoogleTTSService` was emitting two `TTSStoppedFrames`.
+
+### Performance
+
+- `BotSpeakingFrame`s are now sent every 200ms. If the output transport audio chunks
+  are higher than 200ms then they will be sent at every audio chunk.
+
 ### Other
 
 - Added foundational example `37-mem0.py` demonstrating how to use the

src/pipecat/transports/base_output.py

@@ -336,13 +336,22 @@ class BaseOutputTransport(FrameProcessor):
             return without_mixer(BOT_VAD_STOP_SECS)
 
     async def _sink_task_handler(self):
+        # Push a BotSpeakingFrame every 200ms, we don't really need to push it
+        # at every audio chunk. If the audio chunk is bigger than 200ms, push at
+        # every audio chunk.
+        TOTAL_CHUNK_MS = self._params.audio_out_10ms_chunks * 10
+        BOT_SPEAKING_CHUNK_PERIOD = max(int(200 / TOTAL_CHUNK_MS), 1)
+        bot_speaking_counter = 0
         async for frame in self._next_frame():
             # Notify the bot started speaking upstream if necessary and that
             # it's actually speaking.
             if isinstance(frame, TTSAudioRawFrame):
                 await self._bot_started_speaking()
-                await self.push_frame(BotSpeakingFrame())
-                await self.push_frame(BotSpeakingFrame(), FrameDirection.UPSTREAM)
+                if bot_speaking_counter % BOT_SPEAKING_CHUNK_PERIOD == 0:
+                    await self.push_frame(BotSpeakingFrame())
+                    await self.push_frame(BotSpeakingFrame(), FrameDirection.UPSTREAM)
+                    bot_speaking_counter = 0
+                bot_speaking_counter += 1
 
             # No need to push EndFrame, it's pushed from process_frame().
             if isinstance(frame, EndFrame):