Convert audio and voice notes to text you can copy, search and export, without uploading them anywhere. Transcription runs on your own computer with OpenAI's open Whisper models.
- Private. Audio and transcripts never leave your computer. After a one-time model download it works offline.
- Any recording. Opens
.opus,.ogg,.m4a,.mp3,.wav,.amr,.aac,.webmand more. - Sessions. Every batch of notes becomes a thread you can rename, search, add to and delete.
- Many ways in. Pick files, convert a whole folder, drag and drop, or record from your microphone.
- Listen along. Each line has a timestamp; click it to hear that exact moment.
- Get the text out. Copy one note or the whole session, with or without times, or export everything as a zip.
- Your language. English by default; pick another language next to the model when your notes aren't in English.
- Two models. Fast (Whisper small) suits most voice notes; Accurate (large-v3-turbo) is better with heavy accents and noise.
- Command line too.
vscrybe transcribewrites a.txtnext to every voice note in a folder.
Download the latest VScrybe-linux-amd64.deb from
Releases, then:
sudo apt install ./VScrybe-linux-amd64.debOpen VScrybe from your app menu. The first time you transcribe, it downloads the Fast model (about 480 MB).
Uninstall with sudo apt remove vscrybe. Your sessions stay in ~/.local/share/vscrybe until you delete
that folder.
Download VScrybe-linux-amd64.AppImage from Releases, then:
chmod +x VScrybe-linux-amd64.AppImage
./VScrybe-linux-amd64.AppImageDownload VScrybe-windows-x64-setup.exe from Releases and run it.
The installer isn't code-signed yet, so Windows may show "Windows protected your PC": choose More info → Run
anyway. Uninstall from Settings → Apps.
Download the .dmg for your Mac from Releases:
VScrybe-macos-arm64.dmgfor Apple silicon (M1 and later)VScrybe-macos-x64.dmgfor Intel Macs
Open it and drag VScrybe into Applications. The app isn't notarized by Apple yet, so the first launch is blocked: open System Settings → Privacy & Security, scroll down and click Open Anyway next to VScrybe.
To build from source, see CONTRIBUTING.md.
The .deb installs a vscrybe command. On macOS it is /Applications/VScrybe.app/Contents/MacOS/vscrybe, and on
Windows vscrybe.exe in VScrybe's install folder.
vscrybe transcribe ~/Recordings/ # writes a .txt next to each voice note
vscrybe transcribe note.opus --stdout # print instead of writing a file
vscrybe transcribe note.opus -t # [m:ss] timestamps on every line
vscrybe transcribe interview.m4a -m large-v3-turbo # the Accurate model
vscrybe transcribe note.opus -l fr # a language other than EnglishFiles that already have a .txt are skipped unless you pass --force.
| What | Linux | macOS | Windows |
|---|---|---|---|
| Sessions, audio copies, transcripts | ~/.local/share/vscrybe |
~/Library/Application Support/vscrybe |
%APPDATA%\vscrybe |
| Downloaded models | ~/.cache/vscrybe/models |
~/Library/Caches/vscrybe/models |
%LOCALAPPDATA%\vscrybe\models |
| Log file | ~/.local/state/vscrybe/log/vscrybe.log |
~/Library/Caches/vscrybe/log/vscrybe.log |
%LOCALAPPDATA%\vscrybe\log\vscrybe.log |
Your original files are never changed or moved. Deleting a session deletes its copies.
VScrybe is a single Rust program built with Tauri. It runs a small web server on a private local port (axum + MiniJinja templates, with an htmx + Alpine.js interface) and shows it in a native window. Audio is decoded with FFmpeg built into the app and transcribed with CTranslate2, the same engine faster-whisper uses. Nothing listens on the network.
Bug reports and pull requests are welcome. See CONTRIBUTING.md for setup, tests and how to build the desktop app. Found a problem but don't use GitHub? Use About → Report an issue in the app.
MIT. VScrybe bundles and downloads third-party software and models under their own licenses; see THIRD_PARTY_NOTICES.md.