BenAAndrew / Voice-API

API template for deploying tacotron2 voices

Voice-API

Voice service API available at https://voice-cloning-api.herokuapp.com/

Setup

1. Clone this repository
2. Install Python3
3. Run pip install -r requirements.txt
4. Create a folder called data inside the repo
5. Add each voice model to the data folder in the format firstname_lastname.pt (i.e. David_Attenborough.pt)

Request Format

Make a GET request to the root URL with the name of the voice & given text to synthesise. For example: https://voice-cloning-api.herokuapp.com/?name=David Attenborough&text=Some sample text

This returns a JSON response with the URL for the generated audio and graph. For example: {"audio":"results/1234.wav","graph":"results/1234.png"}

These can then be fetched from the given URLS. i.e. https://voice-cloning-api.herokuapp.com/results/1234.wav

Voice list

The list of available voices can be fetched from https://voice-cloning-api.herokuapp.com/voices