Distavo

Website: distavo.com

Drop a recording in a folder, get a tidy Markdown meeting note back.

Distavo is a native macOS menu-bar app that watches a folder for audio/video recordings and automatically turns each new one into a structured Markdown meeting note. It converts the file locally with AVFoundation, transcribes it with a built-in on-device engine (Whisper, a Fast engine for 25 European languages, two Catalan/Spanish models from the Barcelona Supercomputing Center, or an opt-in language pack — community Whisper fine-tunes for Hebrew, Thai, Tamil, Welsh and Norwegian) — or on your own WhisperX server if you prefer — summarises the transcript with your own Ollama server, validates the result, and writes a note to your notes folder.

Distavo does not require any cloud service: the built-in transcription engines run entirely on-device, and for anything else you point Distavo at WhisperX or Ollama endpoints that you run and trust. macOS only.

Screenshots

The native menu and settings window:

Requirements / Prerequisites

• macOS 14+ (this is a menu-bar app — macOS only; this is Distavo's deployment target).
• Apple Silicon Mac for the built-in transcription engines — Intel Macs use a WhisperX server instead (see below). The two Catalan/Spanish models and the large-v3-based language packs (Thai, Tamil, Welsh, Norwegian) additionally need 16 GB of memory or more; the Hebrew pack runs on any Apple Silicon Mac.
• Ollama for summaries — a reachable Ollama HTTP endpoint with a model pulled (for example llama3.1:8b), see Ollama.
• Optional: a reachable WhisperX HTTP endpoint that you run — see WhisperX — if you'd rather transcribe on your own server than use the built-in engines.

An opt-in preview of Apple's on-device Foundation Models summariser (macOS 26) can be enabled with summarise.embedded_enabled in the settings file; Ollama remains the default.

Any server you configure (WhisperX, Ollama) can be on localhost, on another machine on your network, or anywhere you can reach — Distavo never starts it for you, and only ever talks to the URLs you configure.

Install

Distavo is heading to the Mac App Store and Setapp; a notarized direct-download build is published on GitHub Releases.

To build from source you need XcodeGen (brew install xcodegen):

cd apple
xcodegen generate
xcodebuild -project Distavo.xcodeproj -scheme Distavo -configuration Release \
  -derivedDataPath build CODE_SIGNING_ALLOWED=NO build
# the app is written under apple/build/Build/Products/Release/Distavo.app

Distavo launches as a menu-bar agent (no Dock icon) and can start at login from its own settings.

Configure

Open Settings… from the menu bar and fill in:

• your transcription backend — built-in (choose Automatic, or a specific engine and language) or your own WhisperX URL,
• your Ollama URL and model for summarisation,
• the watch / notes / work folders if you want non-default locations.

Use the Test connection button to confirm Distavo can reach WhisperX and Ollama before you drop in a recording.

How it works

1. Distavo watches the recordings folder (default ~/Documents/Distavo/recordings) on a configurable interval.
2. When a new recording appears, AVFoundation converts it locally to WAV.
3. The WAV is transcribed by a built-in on-device engine — or uploaded to your configured WhisperX server, if you use one instead.
4. The cleaned transcript is sent to your configured Ollama model for summarisation.
5. The summary is validated and written as Markdown to the notes folder (default ~/Documents/Distavo/notes/<name>.md).

Config and the work/cache directory live under ~/Library/Application Support/Distavo/, and logs are written to ~/Library/Logs/Distavo/distavo.log.

Supported input formats include .wav, .m4a, .mp3, .opus, .ogg, .flac, .aac, .mov, .mp4, and .m4v (anything AVFoundation can decode).

Privacy

Distavo is built to keep your data on machines you control:

• Audio is converted to WAV locally with AVFoundation.
• With the built-in engines, transcription happens entirely on-device and nothing is uploaded. If you configure a WhisperX server instead, the WAV is uploaded only to that server; the transcript is sent only to the Ollama server you configured. These may be remote, so point Distavo only at servers you trust.
• Notes and transcripts are written in cleartext under your home directory.
• There is no telemetry and no phone-home. Nothing is sent anywhere except the WhisperX/Ollama endpoints you configure — and downloading a built-in model the first time it's needed, from Distavo's own Hugging Face repo.

See PRIVACY.md for the full statement.

Contributing

Contributions are welcome — see CONTRIBUTING.md and the Code of Conduct.

Roadmap

Planned direction (direct download → Setapp → Mac App Store) is described in ROADMAP.md.

License

MIT — see LICENSE.

Support

Distavo costs $29 on the Mac App Store — a one-time purchase, no subscription, all future updates included. That is the convenient way to buy it: sandboxed, auto-updating, and it funds the work.

The app is also MIT-licensed, so the notarized direct DMG on GitHub Releases stays free, and you can always read or build the source yourself. Paying is for convenience and support, not for permission — the point of a tool that promises your meetings never leave your machines is that you can audit it.

If you use the free build and it saves you time, you can support its development:

Tracking

Vikunja project 18 · Distavo — https://familia.riera.co.uk/projects/18