OpenVoice Lab is a Windows desktop app for local text-to-speech (TTS). It pairs a WinUI front-end with a bundled worker that runs entirely on your machine—no cloud account or Python install required.
- Generate speech from text using local TTS models.
- Manage voices, projects, and pronunciation profiles.
- Download and store models on your PC for offline use.
- Windows 11 (x64)
- 10+ GB free disk space for models (more if you keep multiple sizes)
- Optional: NVIDIA GPU with compatible CUDA drivers for faster generation
- Download the latest installer from GitHub Releases:
https://github.com/<OWNER>/<REPO>/releases/latest/download/OpenVoiceLab-Setup.exe
- Run the installer (no admin rights required).
- Launch OpenVoice Lab from the Start menu.
Note: Replace <OWNER>/<REPO> with this repository path if you forked or renamed it. The installer is published as a Release asset and is not stored in the git repo.
By default, the app installs to:
%LOCALAPPDATA%\Programs\OpenVoiceLab
App data (models, voices, logs) is stored in:
%LOCALAPPDATA%\OpenVoiceLab
- Open the app.
- Go to Models and download a model (downloads are on-demand; once a model is downloaded, it can be used offline).
- Go to Playground to generate speech.
The worker runtime starts automatically in the background and uses a local-only connection (127.0.0.1). You never need to open a terminal or manage Python.
- Use Settings → Apps → Installed apps → OpenVoice Lab → Uninstall.
- The uninstaller prompts to remove user data (voices, models, logs) from
%LOCALAPPDATA%\OpenVoiceLab.
- App says worker is unavailable: restart the app to relaunch the worker.
- Find logs: worker and app logs are stored in
%LOCALAPPDATA%\OpenVoiceLab\logs. - Slow generation: ensure you downloaded a model and that your GPU drivers are up to date.
- Disk usage is high: remove unused models from the Models page or delete
%LOCALAPPDATA%\OpenVoiceLab\models\.
See the scripts in scripts/ for building and packaging the app and worker.