I recently set up Voicebox (by jamiepine), a powerful open-source desktop app for voice cloning, text-to-speech, and audio production that runs 100% locally.
Installation Experience:
The native build on Debian worked, but the generated .deb package had issues (missing binary), and the Tauri app kept crashing due to NVIDIA driver compatibility problems with my GeForce GT 1030.
After several attempts with environment variables and GPU workarounds, I switched to the Docker method, which proved much more stable.
Final Working Setup:
- Started with docker compose up -d
- Access the web UI at http://localhost:17600
- Created a simple start script and desktop launcher for easy access
- Running in CPU mode (models download automatically on first use)
First Impressions: Voicebox feels very polished. It supports multiple TTS engines (Qwen3-TTS, Chatterbox, etc.), excellent voice cloning from just a few seconds of audio, a multi-track stories editor, global dictation, and even agent integration.
Verdict: Despite the initial Linux setup hurdles, it’s now running smoothly and is an excellent free/local alternative to paid services like ElevenLabs.
If you’re technical and patient (especially on Linux + NVIDIA), it’s absolutely worth it.