encounter/adk-python
0
0
Fork 0
mirror of https://github.com/encounter/adk-python.git synced 2026-07-09 18:19:28 -07:00
Code Issues Packages Projects Releases Wiki Activity

feat: Move livebidi agents esp multi-agent to use session/events

Browse Source 
The old live/bidi agents are using a cache to store context/history during agent transfer etc. As we have added support for session for live/bidi, we are now migrating the context/history cache to it. This improves scalability, efficiency and maintainability.

It introduces several changes:
* AudioTranscriber support is removed as now we are using native transcription from models.
* Transcription is returned as input_transcription/output_transcription fields and no longer as contents.
* We will return a new event with artifact references of file type of audio/pcm.(in addition to existing audio response event. So the users of this api need to do proper filtering here.)

PiperOrigin-RevId: 805997675
This commit is contained in:
Hangfei Lin
2025-09-11 14:58:33 -07:00
committed by Copybara-Service
parent 873551d7b9
commit ab69ef8de8
12 changed files with 435 additions and 417 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/google/adk/models/gemini_llm_connection.py
+2 -22
View File
@@ -166,38 +166,18 @@ class GeminiLlmConnection(BaseLlmConnection):
message.server_content.input_transcription
and message.server_content.input_transcription.text
):
user_text = message.server_content.input_transcription.text
parts = [
types.Part.from_text(
text=user_text,
)
]
llm_response = LlmResponse(
content=types.Content(role='user', parts=parts)
input_transcription=message.server_content.input_transcription,
)
yield llm_response
if (
message.server_content.output_transcription
and message.server_content.output_transcription.text
):
# TODO: Right now, we just support output_transcription without
# changing interface and data protocol. Later, we can consider to
# support output_transcription as a separate field in LlmResponse.
# Transcription is always considered as partial event
# We rely on other control signals to determine when to yield the
# full text response(turn_complete, interrupted, or tool_call).
text += message.server_content.output_transcription.text
parts = [
types.Part.from_text(
text=message.server_content.output_transcription.text
)
]
llm_response = LlmResponse(
content=types.Content(role='model', parts=parts), partial=True
output_transcription=message.server_content.output_transcription
)
yield llm_response
if message.server_content.turn_complete:
if text:
yield self.__build_full_text_response(text)