🎤

Transcribe audio

dbn.transcribe_audio

Transcribe an audio file from base64-encoded content or a URL.

Parameters

Parameter	Type	Required	Description
audio_base64	string	—	Base64-encoded audio file content. Use either this or audio_url.
audio_url	string	—	URL to a publicly accessible audio file. Use either this or audio_base64.
filename	string	—	Original filename with extension (e.g. recording.mp3) — helps set the correct audio format.
language	string	—	Language spoken in the audio (e.g. no, en, uk). Leave blank for auto-detection.
diarize	boolean	—	Enable speaker diarization — label each segment with a speaker identifier (true/false).

* Purple = required

Example request

POST /api/mcp/user/tools/dbn.transcribe_audio/invoke

{
  "audio_url": "https://example.com/meeting-recording.mp3",
  "language": "no",
  "diarize": true
}

Example response

{
  "ok": true,
  "transcript": "Full transcript of the audio recording...",
  "segments": [
    { "speaker": "A", "start": 0.0, "end": 4.2, "text": "God morgen, vi er samlet for å diskutere..." },
    { "speaker": "B", "start": 4.5, "end": 9.1, "text": "Takk. Jeg ønsker å starte med å si at..." }
  ],
  "language_detected": "no",
  "credits_used": 8
}

Connect

Create your MCP token on the setup page and use it with any supported client.

Set up MCP →

Privacy

Process-and-forget by default. All tool calls process your text in memory and return results to your AI client. Nothing is saved to My Case unless you explicitly call dbn.save_to_case.

Tools provide legal preparation support, not final legal advice. Results are for informational purposes and should be reviewed by a qualified legal professional.