Grok Voice 3.0 with application-owned backend work over WebSocket.
Speakable context uses conversation items and explicit response continuation. Socket submission is unconfirmed; caller transcripts are finalized per item. Quiet/instruction updates, usage reporting and transparent resumption are not advertised. Input mute gates local audio while sending codec silence.
Summary
Functions
Finite receive limits for generated speech bursts, in the selected pipeline format.
Returns a specification to start this module under a supervisor.
Functions
Finite receive limits for generated speech bursts, in the selected pipeline format.
Returns a specification to start this module under a supervisor.
See Supervisor.