Cross-platform desktop app for transcribing audio and video files using local AI models. No cloud, no API keys — everything runs on your machine.
Inspired by Handy — a great open-source speech-to-text app. HandyFiles borrows its model management approach, uses the same transcribe-rs engine, and shares the GigaAM model distribution. While Handy focuses on live microphone recording, HandyFiles is designed for file-based transcription with drag & drop.
- Drag & drop any audio or video file
- Local transcription using Whisper and GigaAM models
- Model management — download, switch, and compare models from the app
- Native audio decoding — no FFmpeg required (Symphonia + Rubato)
- Re-transcribe with a different model to compare results
- Supports: MP3, WAV, FLAC, OGG, AAC, M4A, MP4, MKV, WebM
| Model | Size | Languages | Notes |
|---|---|---|---|
| Whisper Tiny | 75 MB | 99 languages | Fastest |
| Whisper Base | 142 MB | 99 languages | Good balance |
| Whisper Small | 466 MB | 99 languages | Recommended |
| Whisper Medium | 1.5 GB | 99 languages | High quality |
| Whisper Large v3 | 3 GB | 99 languages | Best quality |
| Whisper Large v3 Turbo | 1.6 GB | 99 languages | Large quality, medium speed |
| GigaAM v3 | 151 MB | Russian | Specialized for Russian |
Download the latest release for your platform from Releases.
- macOS:
.dmg(Apple Silicon) - Windows:
.exe(NSIS installer) - Linux:
.debor.AppImage
The app is not signed with an Apple Developer certificate. macOS Gatekeeper will block it. To fix, run in Terminal after installing:
xattr -cr /Applications/HandyFiles.appLinux only:
sudo apt install libwebkit2gtk-4.1-dev libgtk-3-dev libayatana-appindicator3-dev librsvg2-devpnpm install
pnpm tauri buildpnpm tauri dev- Tauri v2 — Rust backend + webview frontend
- React 18 + TypeScript + Tailwind CSS 4
- transcribe-rs — Whisper (whisper-cpp) + GigaAM (ONNX) transcription
- Symphonia — pure Rust audio/video decoding
- Rubato — audio resampling to 16kHz
MIT