Speech-to-Text Streaming

​

Endpoint

​

Authentication

​

Query Parameters

​

Connection example

wss://api.munsit.com/api/v1/websocket/speech-to-text?x-api-key=YOUR_MUNSIT_API_KEY&model=munsit

​

Supported audio input

• First chunk must be WAV (with headers)
• Subsequent chunks can be WAV or raw PCM

​

Client message format

{
  "event": "audio_chunk",
  "data": {
    "audioBuffer": [1, 2, 3]
  }
}

{
  "audioBuffer": [1, 2, 3]
}

​

Events to emit

​

audio_chunk

• Direction: client -> server
• Payload: audioBuffer as Array<Uint8>
• Guidance:
  • send approximately every ~1 second
  • first chunk should include full WAV headers
  • after first chunk, PCM chunks are accepted

​

Events to listen to

​

transcription

• Direction: server -> client
• Type: string
• Meaning: cumulative Arabic transcript generated from all received chunks

​

transcription_error

• Direction: server -> client
• Type: string
• Meaning: error details during streaming transcription

​

Connection/auth notes

• connect: client-level WebSocket open event (successful handshake)
• authentication_error: not emitted as a dedicated WS event in current backend; invalid auth is handled by rejecting/closing the connection

​

Recommended flow

1. Connect to WS /websocket/speech-to-text with auth
2. Confirm the socket is open on the client side
3. Emit audio_chunk payloads (roughly every ~1s)
4. Listen for transcription_error and handle failures
5. Listen for transcription and render live text updates
6. Safely disconnect when transcription is complete