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

Only send TTSStoppedFrame from Gemini when in AUDIO mode, only send one LLMFullResponseEndFrame

Browse Source
This commit is contained in:
Mark Backman
2025-05-02 10:18:53 -04:00
parent 38e2d37674
commit 96d40903a9
1 changed files with 5 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/pipecat/services/gemini_multimodal_live/gemini.py
View File

@@ -905,11 +905,12 @@ class GeminiMultimodalLiveLLMService(LLMService):
text = self._bot_text_buffer text = self._bot_text_buffer
self._bot_text_buffer = "" self._bot_text_buffer = ""
# Pertains to modalities set to TEXT only # Only push the TTSStoppedFrame the bot is outputting audio
if text: # when text is found, modalities is set to TEXT and no audio
await self.push_frame(LLMFullResponseEndFrame()) # is produced.
if not text:
await self.push_frame(TTSStoppedFrame())
await self.push_frame(TTSStoppedFrame())
await self.push_frame(LLMFullResponseEndFrame()) await self.push_frame(LLMFullResponseEndFrame())
async def _handle_evt_output_transcription(self, evt): async def _handle_evt_output_transcription(self, evt):