OmniVoice TTS API

This is a standalone Text-to-Speech API built with FastAPI, based on the powerful OmniVoice engine. It provides a simple REST interface for high-quality, multilingual voice cloning and synthesis.

Credits

This project is based on the original OmniVoice repository by k2-fsa. All credits for the core model architecture and training go to the original authors.

License

Distributed under the Apache 2.0 License. See LICENSE for more information.

Support the Developer

If you find this project useful, consider supporting me:

Or send Solana / USDC / USDT via Solana network:

9rupbyrM19RaVbHmJ4fusozux6P9t72GoYB7Sdy4Nmks

---

Requirements

Before getting started, make sure you have the following installed on your system:

Note: The API can run on CPU, but generation will be significantly slower. A GPU with at least 6GB VRAM is strongly recommended for real-time usage.

---

Installation & Setup

1. API Setup (using uv)

This project uses uv for fast and reliable Python package management. You must have uv installed to run the backend.

1. Install uv: Follow the instructions at astral.sh/uv to install it on your system.

2. Setup Environment: In the root directory of this repository, run:

   uv sync
   

   This will create a virtual environment and install all necessary dependencies (including torch, fastapi, etc.).

3. Run the API: You can use the provided batch file (Windows) or run manually:

   # Using the batch file
   run_api.bat
   
   # Manual run
   uv run uvicorn api:app --host 0.0.0.0 --port 8000 --reload
   

2. Frontend Setup (using npm)

The frontend is built with React and Vite.

1. Navigate to frontend folder:

   cd frontend
   

2. Install dependencies:

   npm install
   

3. Run the frontend:

   npm run dev
   

   The frontend will typically be available at http://localhost:5173.

3. Voice Library Management

You can easily add new voices for cloning through the web interface:

1. Access the Frontend: Open your browser and navigate to the frontend URL (default: http://localhost:5173).
2. Open Voice Library: Click on the Voice Library tab/button.
3. Provide Details:
   • Voice Name: Give your voice a descriptive name.
   • Reference Audio: Select the .wav or .mp3 file you want to clone.
   • Reference Transcript (Optional): Type the exact words spoken in the audio. Leave this empty to let the system automatically transcribe it for you.
4. Save: Click Save voice to library.

To delete a voice: Simply go to the voices/ folder in the root directory and delete the folder named after the voice you want to remove.

---

API Documentation

Once the API is running, you can access the interactive Swagger documentation at:

• http://localhost:8000/docs

For a detailed breakdown of endpoints, see API_DOCS.md.

Features

• Zero-Shot Voice Cloning: Clone any voice with just a few seconds of reference audio.
• Multilingual Support: Supports high-quality synthesis for English, Chinese, Japanese, Korean, and Indonesian.
• FastAPI Backend: High-performance, asynchronous API.
• Modern UI: Clean and responsive web interface for easy testing.