Skip to main content

Message Types

1. Initial Message (Client to Server)​

Purpose​

Establish the transcription session with initial parameters.

Sent​

Once per session, immediately after the websocket connection is established.

Format​

JSON

Fields​

  • multilingual: Indicates if multilingual transcription is enabled (boolean)
  • language: Language code for transcription (string, e.g., "en" for English)
  • task: Task to perform, e.g., "transcribe", "translate" (string)
  • auth: Authentication token or API key (string, optional)

Example request​

{
    "multilingual": false,
    "language": "en",
    "task": "transcribe",
    "auth": "your-auth-token"
}

Example response​

{
    "uid": "2c0b8c68-c708-4d44-9c0e-01b7871458a7",
    "message": "SERVER_READY"
}

2. Audio Data (Client to Server)​

Purpose​

Stream audio data for transcription.

Sent​

Continuously after the initial message, as audio is captured.

Format​

Binary

Content​

Audio data in Float32Array format, resampled to 16kHz.

3. Transcription Results (Server to Client)​

Purpose​

Receive transcribed text from the server.

Format​

JSON

Fields​

  • uid: Version 4 UUID
  • segments: An array of segments containing the start time, end time, and transcribed text:
    • start: float
    • end: float
    • text: string

Example response​

{
 "uid": "2c0b8c68-c708-4d44-9c0e-01b7871458a7",
 "segments": [
   {"start": 0, "end": 1.2, "text": "Hello, how are you?"},
   {"start": 1.5, "end": 3, "text": "I'm good, thank you for asking."}
 ]
}