Discussions
use LiveAvatar with LiveKit in Custom Mode and connect it to Gemini Live
I want to use LiveAvatar with LiveKit in Custom Mode and connect it to Gemini Live.
Do you have an official JavaScript / Node.js plugin or SDK for the LiveAvatar + LiveKit integration or the new LiveKit Avatar Plugin?
And if not, is it possible to build this pipeline using Node.js, or is it only supported in Python right now?
LiveAvatar FULL mode: audio works but no remote participant/video in LiveKit
We’re integrating LiveAvatar FULL mode. /v1/sessions/start returns a livekit_url and livekit_client_token (identity is client). We connect with @livekit/react-native using
that token. Audio plays (we hear the avatar), but remoteParticipants is always empty and no trackSubscribed events fire, so we can’t render video. Logs show no participants
before disconnect. Sample start payload (token redacted):
Can `stream=True` be accepted when calling LLM in custom mode
Hi there,
Typo in stop session API reference
"Stops a running session. If the session token from Create Session Token is used, we do not require a session token."
Typo in Quick Guide
"When successful, you'll receive both a session_id and session_id." should probably read "When successful, you'll receive both a session_id and session_token.".
Change Avatar Background
Does liveavatar currently support changing the background in stock and custom created avatars?
livkit integration issue
first question is in fullmode im getting livkit url and token from which i can join the room and chat with avatar then how do i record the session like whole screen.
second question im using full mode can i pass my own knowledge base if yes then how is context is the same as knowledge base?
Why there is not livekit_agent_token?
Should my backend also join the livekit room with the same livekit_client_token?
API returned error: { code: 5000, data: null, message: 'Internal server error' }
hi
New usecase
I wanna use liveavatar with livekit in Custom mode where i wanna use Gemini live model for livekit to handle all the input and give the audio response to the avatar to speak it