DhivehiTTS-http

Simple FastAPI app to server Dhivehi.ai pretrained TTS models.

Setup

1. git clone https://github.com/ashraq1455/DhivehiTTS-http.git
2. Download dhivehi.ai pre-trained models from here
3. Extract and copy the models to models/female/ and models/male/ directory
4. pip install -r requirements.txt
5. python main.py to run development server

Directory Structure

|- models/
    |- female/
      |- tts.saved_model
      |- vocoder.saved_model
    |- male/ 
      |- tts.saved_model
      |- vocoder.saved_model
    |- __init__.py
    |- tts.py
|- main.py

How to use

Send a GET request to http://localhost:8000/predict/?text=ހެލޯ ދިވެހި&voice=female&speed=0.9

Running on Docker

There's a simple Dockerfile included to setup a quick deployment for testing to use,

1. 'docker build --tag tts-dv .'
2. 'docker run -it --rm -p 8000:8000 tts-dv'