A cross-platform desktop application for Text-to-Speech using Qwen3-TTS models with voice cloning, custom voices, and voice design capabilities.

• Voice Cloning: Clone any voice from a reference audio sample
• TTS Custom Voice: Generate speech using preset voice models
• Voice Design: Create custom voices by adjusting age, gender, accent, and emotion
• Auto Model Download: Automatically downloads Qwen3-TTS models from HuggingFace
• Cross-Platform: Works on Windows, macOS, and Linux
• Dark/Light Mode: Comfortable interface for any lighting condition

• Frontend: Electron + React + TypeScript + Tailwind CSS
• Backend: Python + FastAPI + PyTorch
• Models: Qwen3-TTS-1.7B from HuggingFace

• Node.js 18+ and npm
• Python 3.9+
• Git
• 8GB+ RAM recommended
• 10GB+ disk space for models
• GPU optional but recommended (CUDA-compatible)

cd c:\Development\GemTTS

cd backend

# Create virtual environment
python -m venv venv

# Activate virtual environment
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
# source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

cd ../frontend

# Install dependencies
npm install

Terminal 1 - Backend:

cd backend
venv\Scripts\activate  # On Windows
python main.py

Terminal 2 - Frontend:

cd frontend
npm run electron:dev

cd frontend
npm run electron:dev

(This will automatically start the Python backend)

cd frontend
npm run electron:build:win   # For Windows
npm run electron:build:mac   # For macOS
npm run electron:build:linux # For Linux

The built application will be in frontend/dist-electron/

1. First Launch: The application will prompt you to download the Qwen3-TTS models. This is a one-time setup that may take 10-30 minutes depending on your internet connection.

2. Voice Cloning Tab:

   • Upload a reference audio file (WAV, MP3, etc.)
   • Enter the text you want to speak
   • Adjust similarity and speed parameters
   • Click "Generate Voice"

3. TTS Custom Voice Tab:

   • Select a preset voice from the dropdown
   • Enter your text
   • Adjust speed and pitch
   • Click "Generate Speech"

4. Voice Design Tab:

   • Enter your text
   • Adjust age, gender, accent, and emotion sliders
   • Save/load presets for later use
   • Click "Generate Voice"

GemTTS/
├── backend/
│   ├── main.py              # FastAPI server
│   ├── model_manager.py     # Model download and management
│   ├── tts_processor.py     # TTS inference logic
│   ├── requirements.txt     # Python dependencies
│   ├── models/              # Downloaded models (auto-created)
│   ├── uploads/             # Uploaded audio files (auto-created)
│   └── outputs/             # Generated audio (auto-created)
├── frontend/
│   ├── electron/
│   │   ├── main.js          # Electron main process
│   │   └── preload.js       # Electron preload script
│   ├── src/
│   │   ├── components/      # React components
│   │   │   ├── VoiceCloning.tsx
│   │   │   ├── TTSCustomVoice.tsx
│   │   │   ├── VoiceDesign.tsx
│   │   │   └── ModelsStatus.tsx
│   │   ├── api/
│   │   │   └── apiService.ts # API client
│   │   ├── App.tsx          # Main app component
│   │   ├── App.css          # Styles
│   │   └── main.tsx         # Entry point
│   ├── package.json
│   ├── vite.config.ts
│   └── tsconfig.json
├── README.md
└── LICENSE

Edit backend/main.py to change:

• API host/port (default: 127.0.0.1:8000)
• Model paths
• Output settings

Edit frontend/src/api/apiService.ts to change:

• API endpoint URL
• Request timeouts

• Check your internet connection
• Ensure you have enough disk space (10GB+)
• Check HuggingFace is accessible from your network

• Ensure models are fully downloaded
• Check Python backend logs in the terminal
• Verify your system has enough RAM (8GB+ recommended)

• Install CUDA toolkit (11.8+)
• Install PyTorch with CUDA support:

  pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

• Ensure Python backend is running first
• Check that port 8000 is not in use
• Look for errors in the terminal

1. Backend: Add new endpoints in main.py and processing logic in tts_processor.py
2. Frontend: Create new components in src/components/ and wire them up in App.tsx

# Backend
cd backend
pytest

# Frontend
cd frontend
npm test

MIT License - see LICENSE file for details

• Qwen3-TTS by Alibaba Cloud
• Built with Electron, React, FastAPI, and PyTorch

For issues and questions, please open an issue on GitHub.

• Batch processing multiple texts
• Voice preset library with community voices
• Real-time voice morphing
• SSML support for advanced text markup
• Multi-language support
• Voice fine-tuning interface
• Audio effects and post-processing
• Export to multiple formats