climatologist/f5-tts repository overview

⁠F5-TTS Docker Image

A containerized implementation of F5-TTS (Flow Matching for Zero-Shot Text-to-Speech) optimized for GPU-accelerated environments.

⁠Quick Start

docker pull climatologist/f5-tts:latest
docker run -p 8000:8000 --gpus all climatologist/f5-tts:latest

⁠Features

• 🚀 GPU Accelerated: CUDA 12.1 support with PyTorch 2.7.1
• 🎯 Production Ready: FastAPI-based REST API
• 🔧 Plug & Play: Pre-configured with all dependencies
• 📦 Self-Contained: Automatic model downloading on first run

⁠API Endpoints

• GET /health - Health check and GPU status
• POST /synthesize - Text-to-speech synthesis
• GET / - API documentation

⁠System Requirements

• GPU: NVIDIA GPU with CUDA support
• Memory: 8GB+ GPU VRAM recommended
• CUDA: 12.1+ compatible GPU required
• Storage: 20GB+ container disk space

⁠Usage

⁠RunPod Deployment

• Container Image: climatologist/f5-tts:latest
• Volume Mount Path: /workspace (for persistent data)
• Expose Port: 8000
• Start Command: Leave blank (it's handled by the image's ENTRYPOINT).

⁠Local Testing

To run locally while persisting your data (e.g., models, outputs), use a volume mount:

docker run -p 8000:8000 --gpus all -v $(pwd)/my_tts_data:/workspace climatologist/f5-tts:latest

Access the API at http://localhost:8000

⁠Tags

• latest - Current stable release
• Built on pytorch/pytorch:2.4.0-cuda12.1-cudnn9-runtime with PyTorch 2.7.1

⁠Support

For technical issues, please refer to the project documentation.

Image maintained by climatologist