Plaud recipe

Plaud is an acquisition example, not a core dependency.

This page is the short, technical form. For an end-to-end walkthrough aimed at non-developers, see Transcribe your Plaud recordings on Ubuntu.

1. Use an authorized Plaud export, CLI, connector, or MCP integration to list recordings and obtain an original-audio URL.
2. Pass that temporary HTTPS URL to transcribe_audio.
3. Consume the canonical result, not Plaud- or provider-specific fields.
4. If stored mode was used, read all chunks and call delete_transcript.

{
  "request": {
    "source": {"type": "url", "url": "https://AUTHORIZED-TEMPORARY-URL"},
    "provider": "auto",
    "routing_policy": "quality",
    "diarization": true,
    "timestamps": "segment",
    "transcript_style": "clean",
    "source_delivery": "auto",
    "result_mode": "auto"
  }
}

This request states diarization, timestamps, and transcript_style because a multi-speaker meeting needs them: a stated capability excludes every model that cannot honour it, so this example will not route to Groq Whisper. Drop the three fields to let the request run on any configured provider and read back what was applied from the response metadata.

Vendor connector fields and URL lifetimes can change and are outside the OpenTranscribe compatibility contract. Do not extract credentials, reverse engineer protected systems, bypass access controls, or use audio without the necessary rights.

Plaud is a trademark of its respective owner. OpenTranscribe is independent and is not affiliated with, endorsed by, or sponsored by Plaud.