This project aims to provide a Hindi language Text-to-Speech (TTS) system using code from VITS and a model from Facebook. The system allows users to input text and receive an audio stream as output through a WebSocket interface.

• Introduction
• Setup
• Usage
• License

The Hindi Language Text-to-Speech (TTS) project utilizes code from VITS (Variable Identity Transformer Synthesis) and a model developed by Facebook. TTS technology converts written text into synthesized speech, enabling applications to generate human-like speech from input text.

This project specifically focuses on providing a TTS system for the Hindi language, allowing users to convert written Hindi text into high-quality speech output. By leveraging the power of deep learning and natural language processing, the project aims to provide an accurate and expressive TTS solution for Hindi speakers.

We suggest you to use docker for convience:

1. Clone the repository:

   git clone https://github.com/haoxingxing/TTSAudioWS
   cd TTSAudioWS
   

2. Docker build

   docker build . -t TTSAudioWS
   

or you can install it directly (Tested in ubuntu 22.04)

1. Clone the repository:

   git clone https://github.com/haoxingxing/TTSAudioWS
   cd TTSAudioWS
   

2. Install the required dependencies by running the following command:

   pip install -r requirements.txt
   

3. Install Vits and the facebook model

   ./install_tts.sh
   

4. Start the WebSocket server by running the following command:

   cd vits
   python server.py
   

After you start the server,it will listen on 0.0.0.0:8765
The api follows the doc in https://support.huaweicloud.com/api-sis/sis_03_0113.html
BUT THE ENDPOINT NEEDN'T TO BE FILLED IN THIS DEMO

We added a demo client client.py you can modify the txt in the py script to test

MIT License