Hi - I've been back and forth with the docs (and l...
# docs-feedbackp
powerful-magician-63563
12/28/2025, 12:03 PMHi - I've been back and forth with the docs (and lk assistant!) for gemini realtime plug-in. I really want a clean push-to-talk pipeline for gemini live api but it wasn't as straightforward as i expected. Here's my approach (in the thread) which is working ok.
I was wondering if anyone could verify it's the right approach? [Gemini doesn't seem to have the same server-side buffer that openai has. openai push to talk was very simple].
powerful-magician-63563
12/28/2025, 12:04 PMSummary: Push-to-talk implementation with Gemini Realtime + LiveKit Agents
What I'm doing:
We're implementing push-to-talk for an iOS app using LiveKit Agents with Gemini Realtime (gemini-2.5-flash-native-audio-preview-09-2025). The user holds a button to
speak, and the agent only responds when they release it.
Configuration:
- RealtimeInputConfig(automatic_activity_detection=AutomaticActivityDetection(disabled=True))
- AgentSession(turn_detection="manual")
- iOS sends RPC calls (start_turn, end_turn) to control turn-taking
Implementation:
- start_turn: Enable audio input, call rt_session.start_user_activity() to send activity_start to Gemini
- end_turn: Disable audio input, wait ~100ms, then send activity_end via
rt_session._send_client_event(types.LiveClientRealtimeInput(activity_end=types.ActivityEnd()))
Question:
We're using _send_client_event() to send activity_end because we couldn't find a public end_user_activity() method (unlike start_user_activity() which is public). Is
this the recommended approach, or is there a public API we should use instead?
d
dry-musician-70452
01/02/2026, 7:02 AMLet me look into this for you!
🙌 1