A realtime audio session. Events follow the OAI realtime schema, same as the /v1/realtime route of llama-server. Inference runs while events() is being read, so keep reading it to get a response.
/v1/realtime
events()
Same as sending input_audio_buffer.append.
input_audio_buffer.append
mono PCM at LoadedContextInfo.audio_input_rate
LoadedContextInfo.audio_input_rate
End the session and free its slot.
Read the server events. It ends when the session ends. Only one reader at a time.
Send a client event, for example session.update or response.create.
session.update
response.create
A realtime audio session. Events follow the OAI realtime schema, same as the
/v1/realtimeroute of llama-server. Inference runs whileevents()is being read, so keep reading it to get a response.