ChatTTS-OpenAI-API

The ChatTTS-OpenAI-API is a project built upon the ChatTTS framework, implementing the v1/audio/speech endpoint in compliance with OpenAI protocols to transform text into speech.

Requirements

• Python 3.8+
• FastAPI
• Pydantic

  Getting Started

  1. Clone the repository
  2. Install the requirements with pip install -r requirements.txt
  3. Run the server with uvicorn main:app --reload

     Usage

     The main endpoint is /v1/audio/speech, which takes a POST request with the following parameters:

• model: (required) The model to use for the speech synthesis.
• input: (required) The text to be transformed into speech.
• voice: (optional) The voice to use for the speech synthesis. Default is 'alloy'.
• response_format: (optional) The format of the response audio file. Default is 'wav'.
• speed: (optional) The speed of the speech. Default is 1.0.
• temperature: (optional) The randomness of the speech. Default is 0.3.
• prompt: (optional) The prompt to use for the speech synthesis. Default is '[oral_2][laugh_0][break_6]'. The voice parameter can take the following values: 'alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'. If an unrecognized voice is provided, 'alloy' will be used. Example request:

  {
  "model": "model_name",
  "input": "Hello, world!",
  "voice": "nova"
  }

  This will return an audio file with the synthesized speech.

  Error Handling

  If a required parameter is missing, a 400 error will be returned. If an error occurs during the speech synthesis process, a 500 error will be returned.

Docker Deployment

This project can be easily deployed using Docker. The Dockerfile included in the repository sets up a Python environment with all the necessary dependencies and sets up the application to run on port 5001.

Building the Docker Image

To build the Docker image, navigate to the project directory and run the following command:

docker build -t chattts-openai-api .

This will create a Docker image named "chattts-openai-api".

Running the Docker Image

Once the Docker image has been built, you can run it with the following command:

docker run -p 5001:5001 chattts-openai-api

This will start the application and bind it to port 5001 on your machine. Please keep in mind that the Docker container must have access to an NVIDIA GPU due to the CUDA dependency of the project.