VoiceCloner

Clone your voice locally with Qwen3-TTS. No data leaves your machine.

Record a few seconds of your voice, then generate speech in your voice from any text. Includes a full audiobook production mode for long-form content.

Features

• Quick Clone -- Record or upload 5-10 seconds of your voice, type text, get speech output
• Audiobook Mode -- Load an EPUB or paste text, chunk it into chapters, generate and review audio per-chunk, export a stitched audiobook
• Voice Library -- Save and manage multiple cloned voices
• VoiceDesign -- Optional style descriptions to control tone and delivery
• Export -- WAV, MP3, FLAC, M4A output formats
• Runs locally -- All inference happens on your machine (MPS/CUDA/CPU)

Quick Start (Python)

Requires Python 3.12+.

git clone https://github.com/thegian7/voicecloner.git
cd voicecloner
./start.sh

start.sh creates a virtual environment, installs dependencies, and launches the Gradio UI at http://localhost:7860.

Or manually:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python app.py

Desktop App (Electron)

The Electron shell wraps the Gradio UI into a native desktop app with automatic Python/venv setup and GPU detection.

npm install
npm run dist:mac    # or dist:win / dist:linux

This bundles a standalone Python runtime (downloaded via npm run download-python) so end users don't need Python installed.

Build targets

Platform	Format	GPU Support
macOS	DMG (arm64 + x64)	MPS (Apple Silicon)
Windows	NSIS installer (x64)	CUDA, CPU
Linux	AppImage (x64)	CUDA, ROCm, CPU

TTS API Server

tts_server.py exposes an OpenAI-compatible /v1/audio/speech endpoint using your cloned voice. Useful for integrating with other tools.

source .venv/bin/activate
python tts_server.py --voice "MyVoice" --port 8765

Project Structure

app.py                  # Main Gradio app (Quick Clone + Audiobook + Voices)
core/                   # Shared TTS engine (model loading, generation, audio processing)
audiobook/              # Audiobook-specific logic (chapters, export, state)
electron/               # Electron desktop shell
python/                 # Python files bundled into Electron builds
tts_server.py           # Standalone TTS API server
start.sh                # One-command launcher

Hardware Requirements

• Apple Silicon Mac: Works out of the box via MPS. 16GB RAM recommended.
• NVIDIA GPU: CUDA 12.4+ with 8GB+ VRAM recommended.
• AMD GPU: ROCm 6.2+ supported.
• CPU: Works but slow. 32GB RAM recommended.

Models are downloaded from Hugging Face on first run (~2GB).

License

MIT