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

transports(base_output): fix constant bot started/stopped speaking frames

Browse Source
This commit is contained in:
Aleix Conchillo Flaqué
2024-10-25 12:10:24 -07:00
parent fcfc729e83
commit 5ab00e01aa
2 changed files with 37 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
CHANGELOG.md
View File

@@ -24,6 +24,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
### Fixed ### Fixed
- Fixed an issue that was generating constant bot started/stopped speaking
frames for HTTP TTS services.
- Fixed an issue that was causing stuttering with AWS TTS service. - Fixed an issue that was causing stuttering with AWS TTS service.
## [0.0.47] - 2024-10-22 ## [0.0.47] - 2024-10-22

42
src/pipecat/transports/base_output.py
View File

@@ -70,9 +70,14 @@ class BaseOutputTransport(FrameProcessor):
self._stopped_event = asyncio.Event() self._stopped_event = asyncio.Event()
# Indicates if the bot is currently speaking. This is useful when we # Indicates if the bot is currently speaking. This is useful when we
# have an interruption since all the queued messages will be thrown # have an interruption since all the queued messages will be thrown away
# away and we would lose the TTSStoppedFrame. # and we would lose the TTSStoppedFrame. We also keep a scheduler handle
# which is used to schedule pushing bot stopped speaking frames. With
# some services (HTTP TTS services) we will get a bunch of
# TTSStartedFrame and TTSStoppedFrame and we don't want to constantly
# generate bot started/stopped speaking frames.
self._bot_speaking = False self._bot_speaking = False
self._bot_stopped_speaking_handle = None
# Create sink frame task. This is the task that will actually write # Create sink frame task. This is the task that will actually write
# audio or video frames. We write audio/video in a task so we can keep # audio or video frames. We write audio/video in a task so we can keep
@@ -320,14 +325,35 @@ class BaseOutputTransport(FrameProcessor):
logger.exception(f"{self} error processing sink clock queue: {e}") logger.exception(f"{self} error processing sink clock queue: {e}")
async def _bot_started_speaking(self): async def _bot_started_speaking(self):
logger.debug("Bot started speaking") # If we scheduled bot stopped speaking, cancel it.
self._bot_speaking = True if self._bot_stopped_speaking_handle:
await self.push_frame(BotStartedSpeakingFrame(), FrameDirection.UPSTREAM) self._bot_stopped_speaking_handle.cancel()
self._bot_stopped_speaking_handle = None
if not self._bot_speaking:
logger.debug("Bot started speaking")
await self.push_frame(BotStartedSpeakingFrame(), FrameDirection.UPSTREAM)
self._bot_speaking = True
async def _bot_stopped_speaking(self): async def _bot_stopped_speaking(self):
logger.debug("Bot stopped speaking") async def push_bot_stopped_speaking_async():
self._bot_speaking = False try:
await self.push_frame(BotStoppedSpeakingFrame(), FrameDirection.UPSTREAM) logger.debug("Bot stopped speaking")
await self.push_frame(BotStoppedSpeakingFrame(), FrameDirection.UPSTREAM)
self._bot_speaking = False
except asyncio.CancelledError:
logger.debug("We've been cancelled")
pass
def push_bot_stopped_speaking():
self.get_event_loop().create_task(push_bot_stopped_speaking_async())
# Schedule a bot stopped speaking to be pushed in half a second (unless
# we get another bot started speaking).
if not self._bot_stopped_speaking_handle:
self._bot_stopped_speaking_handle = self.get_event_loop().call_later(
0.5, push_bot_stopped_speaking
)
# #
# Camera out # Camera out