Diariz
A self-hosted meeting transcription app identifies speakers and chats across recordings using local or cloud OpenAI-compatible models under AGPLv3.

Diariz is a self-hosted meeting transcription app for individuals and teams that want recordings and transcripts on their own server. It turns captured or uploaded audio into searchable meeting records.
- Transcription and speaker recognition. WhisperX large-v3 and pyannote produce editable, playable segments with word-level timestamps, and enrolled voices let you identify recurring speakers across recordings.
- Meeting summaries and actions. You can generate editable minutes from reusable templates and track extracted action items with an owner, deadline and completion status.
- Chat and search. Ask questions across your meeting library with answers linked to exact transcript moments, or combine keyword search with optional semantic embeddings.
- Calendar connections. Google Calendar and public iCal feeds link recordings to events, while recurring meetings show earlier recordings.
- Shared access and Claude. Role-based access and per-user data isolation control access to meeting records; a built-in MCP server lets Claude use your transcripts.
The server runs in Docker and has a browser interface for recording and uploading audio. A Windows desktop app also captures microphone and system audio. Transcription and speaker separation run locally on CPU or GPU.
Summaries and chat require an OpenAI-compatible endpoint, including local models served by Ollama, LM Studio or vLLM. Choosing a cloud endpoint sends the content needed for those requests to that service; connecting Claude also makes selected transcript content available to it.
Diariz is open source under AGPLv3 and free to self-host under that licence. For-profit use without copyleft obligations requires a commercial licence.