Kyutai Tts Docker

🚀 Kyutai TTS Docker Deployment

This project offers a production-ready Docker deployment for Kyutai TTS, featuring a user interface, REST API, and MCP support.

English | 简体中文 | 繁體中文 | 日本語

✨ Features

• 🚀 One-Click Deployment: Automatically selects the GPU and detects ports.
• 🎨 Three Access Modes: Web UI, REST API, and MCP tools are available.
• 🧠 Smart GPU Management: Implements lazy loading and automatic memory release.
• 🌐 Multi-language UI: Offers both English and Chinese interfaces.
• 📦 All-in-One Image: Contains all necessary models without external dependencies.
• 🔒 Production Ready: Comes with HTTPS, health checks, and monitoring features.

🚀 Quick Start

Using Docker Hub (Recommended)

docker run -d \
  --name kyutai-tts \
  --gpus all \
  -p 8900:8900 \
  -e NVIDIA_VISIBLE_DEVICES=0 \
  neosun/kyutai-tts:allinone

Access the application at: http://localhost:8900

Using Docker Compose

git clone https://github.com/neosun100/kyutai-tts-docker.git
cd kyutai-tts-docker
./start.sh

📦 Installation

Prerequisites

• Docker 20.10 or higher
• Docker Compose 2.0 or higher
• NVIDIA GPU with CUDA support
• nvidia-docker runtime

Method 1: Pull from Docker Hub

docker pull neosun/kyutai-tts:allinone

Method 2: Build from Source

git clone https://github.com/neosun100/kyutai-tts-docker.git
cd kyutai-tts-docker
docker-compose build

⚙️ Configuration

Environment Variables

Example .env File

PORT=8900
DEVICE=cuda
GPU_IDLE_TIMEOUT=60
NVIDIA_VISIBLE_DEVICES=0

💻 Usage Examples

Basic Usage

Web UI

1. Open your browser and navigate to http://localhost:8900.
2. Enter the text you want to synthesize.
3. Optionally, adjust the parameters.
4. Click the "Generate" button.
5. You can play or download the generated audio.

REST API

Generate Speech

curl -X POST http://localhost:8900/api/tts \
  -F "text=Hello, world!" \
  -F "cfg_coef=2.0" \
  --output output.wav

Check GPU Status

curl http://localhost:8900/api/gpu/status

Release GPU Memory

curl -X POST http://localhost:8900/api/gpu/offload

MCP Tools

For detailed usage of MCP, refer to MCP_GUIDE.md.

result = await mcp_client.call_tool(
    "text_to_speech",
    {
        "text": "Hello from MCP!",
        "output_path": "/tmp/output.wav"
    }
)

📚 Documentation

Project Structure

kyutai-tts-docker/
├── app.py                 # Flask application
├── gpu_manager.py         # GPU resource manager
├── mcp_server.py          # MCP server
├── Dockerfile             # Docker image
├── Dockerfile.allinone    # All-in-one image
├── docker-compose.yml     # Docker Compose config
├── start.sh               # One-click startup script
├── test_api.sh            # API test script
└── docs/                  # Documentation
    ├── QUICKSTART.md
    ├── MCP_GUIDE.md
    └── TEST_REPORT.md

Tech Stack

• Framework: Flask 3.0
• ML Framework: PyTorch 2.7 + CUDA 12.1
• TTS Model: Kyutai TTS 1.6B (English/French)
• API Docs: Swagger/Flasgger
• MCP: FastMCP 0.2
• Container: Docker + nvidia-docker

API Documentation

Once the application is running, you can access the Swagger documentation at: http://localhost:8900/apidocs

Available Endpoints

• GET /health: Health check
• GET /api/gpu/status: Check GPU status
• POST /api/tts: Generate speech
• POST /api/gpu/offload: Release GPU memory

Production Deployment

With Nginx Reverse Proxy

server {
    listen 443 ssl;
    server_name your-domain.com;
    
    location / {
        proxy_pass http://localhost:8900;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

Multi-GPU Setup

# GPU 0
NVIDIA_VISIBLE_DEVICES=0 PORT=8900 docker-compose up -d

# GPU 1
NVIDIA_VISIBLE_DEVICES=1 PORT=8901 docker-compose up -d

Performance

• Model Size: 1.6B parameters
• GPU Memory: 3 - 4GB
• Latency: 350ms (L40S, 32 concurrent)
• Speed: 3 - 5x real-time
• Audio Quality: 16-bit PCM, 24kHz

🤝 Contributing

Contributions are welcome! Please follow these steps to contribute:

1. Fork the repository.
2. Create a new feature branch (git checkout -b feature/AmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push to the branch (git push origin feature/AmazingFeature).
5. Open a Pull Request.

📝 Changelog

v1.0.0 (2025-12-14)

• Initial release
• Docker deployment with GPU support
• Web UI with multi-language support
• REST API with Swagger docs
• MCP server implementation
• All-in-one Docker image

📄 License

• Python code is licensed under the MIT License.
• Rust code is licensed under the Apache License.
• Model weights are licensed under CC-BY 4.0.

🙏 Acknowledgments

• Thanks to Kyutai Labs for providing the TTS model.
• Thanks to Moshi for the implementation.

⭐ Star History

📱 Follow Us