This project is under early development. APIs, features, and file formats may change without notice.

Audist is a macOS desktop app that records meetings and transcribes them locally using whisper.cpp. No audio or transcripts leave your machine.

Features

• Records microphone and system audio simultaneously
• Transcribes recordings locally using whisper.cpp (base.en model)
• Summarises transcripts via OpenAI, Anthropic, or any OpenAI-compatible provider (e.g. Ollama)
• Rename, browse, and keyboard-navigate past sessions
• Onboarding wizard guides first-time setup
• Preferences window for save directory, LLM provider, and model selection

Tech stack

• Electron + React 19 + TypeScript
• Vite via electron-vite
• Tailwind CSS v4
• whisper.cpp via @remotion/install-whisper-cpp
• ffmpeg-static for audio mixing

Platform support

Audist currently runs on macOS. Windows and Linux support is on the roadmap — contributors are needed to help make it happen. If you're interested, check out CONTRIBUTING.md to get started.

Testers on all platforms are welcome. If you run into any issues, please file a GitHub issue with your platform details and steps to reproduce.

Requirements

• macOS (primary platform — Windows/Linux support is planned)
• Node.js (see .tool-versions for the exact version)
• Xcode Command Line Tools (for building whisper.cpp in development)

Getting started

npm install
npm run dev

On first launch, Audist will guide you through a setup wizard to configure microphone/screen permissions and download the whisper.cpp base model (~140 MB). This only happens once.

Building for release

# Compile the whisper.cpp binary for the current platform (run once before packaging)
npm run prepare:whisper

# Build and package for macOS
npm run build:mac

Testing

# Run e2e tests (Playwright + Electron)
npm run test:e2e

License

Audist is open source software licensed under the GNU General Public License v3. See LICENSE.md for details.