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

better interruption handling by moving the processors after the transport output

Browse Source
This commit is contained in:
Adrian Cowham
2024-10-11 10:33:12 -07:00
parent 678e87fd31
commit 522d931950
3 changed files with 3 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/canonical-metrics/bot.py
View File

@@ -119,9 +119,9 @@ async def main():
user_response, user_response,
llm, llm,
tts, tts,
transport.output(),
audio_buffer_processor, # captures audio into a buffer audio_buffer_processor, # captures audio into a buffer
canonical, # uploads audio buffer to Canonical AI for metrics canonical, # uploads audio buffer to Canonical AI for metrics
transport.output(),
assistant_response, assistant_response,
]) ])

2
examples/chatbot-audio-recording/bot.py
View File

@@ -103,8 +103,8 @@ async def main():
user_response, user_response,
llm, llm,
tts, tts,
audiobuffer, # used to buffer the audio in the pipeline
transport.output(), transport.output(),
audiobuffer, # used to buffer the audio in the pipeline
assistant_response, assistant_response,
]) ])

12
src/pipecat/processors/audio/audio_buffer_processor.py
View File

@@ -33,8 +33,6 @@ class AudioBufferProcessor(FrameProcessor):
self._assistant_audio_buffer = bytearray() self._assistant_audio_buffer = bytearray()
self._num_channels = None self._num_channels = None
self._sample_rate = None self._sample_rate = None
self._assistant_audio = False
self._user_audio = False
def _buffer_has_audio(self, buffer: bytearray): def _buffer_has_audio(self, buffer: bytearray):
return ( return (
@@ -87,14 +85,6 @@ class AudioBufferProcessor(FrameProcessor):
if (isinstance(frame, AudioRawFrame) and self._sample_rate is None): if (isinstance(frame, AudioRawFrame) and self._sample_rate is None):
self._sample_rate = frame.sample_rate self._sample_rate = frame.sample_rate
if isinstance(frame, BotStartedSpeakingFrame):
self._assistant_audio = True
# this handles the case where the user starts speaking and interrupts the bot
if (isinstance(frame, BotStoppedSpeakingFrame) or
isinstance(frame, UserStartedSpeakingFrame)):
self._assistant_audio = False
# include all audio from the user # include all audio from the user
if isinstance(frame, InputAudioRawFrame): if isinstance(frame, InputAudioRawFrame):
self._user_audio_buffer.extend(frame.audio) self._user_audio_buffer.extend(frame.audio)
@@ -105,7 +95,7 @@ class AudioBufferProcessor(FrameProcessor):
self._assistant_audio_buffer.extend(silence) self._assistant_audio_buffer.extend(silence)
# if the assistant is speaking, include all audio from the assistant, # if the assistant is speaking, include all audio from the assistant,
if (isinstance(frame, OutputAudioRawFrame)) and self._assistant_audio: if (isinstance(frame, OutputAudioRawFrame)):
self._assistant_audio_buffer.extend(frame.audio) self._assistant_audio_buffer.extend(frame.audio)
# do not push the user's audio frame, doing so will result in echo # do not push the user's audio frame, doing so will result in echo