"<binary pcm_f32le 100ms audio chunk>"{
"type": "end_of_stream"
}{
"type": "diarization_speaker_start",
"data": {
"timestamp": 1.24,
"speaker": "SPEAKER_00"
}
}{
"type": "diarization_speaker_end",
"data": {
"timestamp": 3.863,
"speaker": "SPEAKER_00"
}
}{
"type": "error",
"message": "Invalid chunk size"
}Documentation Index
Fetch the complete documentation index at: https://docs.pyannote.ai/llms.txt
Use this file to discover all available pages before exploring further.
"<binary pcm_f32le 100ms audio chunk>"{
"type": "end_of_stream"
}{
"type": "diarization_speaker_start",
"data": {
"timestamp": 1.24,
"speaker": "SPEAKER_00"
}
}{
"type": "diarization_speaker_end",
"data": {
"timestamp": 3.863,
"speaker": "SPEAKER_00"
}
}{
"type": "error",
"message": "Invalid chunk size"
}The token is embedded in the WebSocket URL returned by the POST /v1/live REST endpoint. It is single-use and grants access only to this stream session — it carries no team credentials or API key.
Single-use temporary token generated when creating a stream session via POST /v1/live.
Raw PCM audio chunk (binary frame).
End-of-stream signal.
Emitted when a speaker begins a turn.
Emitted when a speaker's turn ends.
Emitted when the server encounters a problem processing a frame.
Was this page helpful?