Event flow sequence
A typical conversation follows this event sequence:
-
Session Start - Initialize the conversation session
-
System Prompt - Send system instructions
-
Chat History (optional) - Provide conversation context
-
Audio Chunks - Stream user audio input
-
Completion Start - AI begins processing
-
ASR Transcripts (USER) - User speech transcription
-
Tool Use (optional) - AI requests tool execution
-
Tool Handling (optional) - Process and return tool results
-
Transcript (ASSISTANT) - SPECULATIVE - Preliminary AI response
-
Audio Chunks - Stream AI audio output
-
Transcript (ASSISTANT) - FINAL - Final AI response transcript
-
Content End Audio - Marks the end of audio content
-
Prompt End - Indicates the completion of the prompt processing
-
Session End - Close the conversation