/Text2SpeechService

Text2Speech service with predictive, dynamic load balancing

Primary LanguagePython

Text to Speech Service

A text-to-speech service implementation with predictive, dynamic load balancing

Setup/Requirements

  1. Install MySQl and run the queries in database/create_tables.sql

  2. Create file database/mysql_credentials.py with following details

     user=<MYSQL_USERNAME>
 host='localhost'    
 password=<MYSQL_PASSWORD>
 database='text2speech'

  3. Install docker and docker-compose and run the following. This will spawn 3 worker nodes and 1 master node, each node being a Gunicorn server process.

     docker-compose build
 docker-compose up

Text2Speech implementation

This repo makes use of FastSpeech for speech synthesis. Download the FastSpeech model into services/text2speech/FastSpeech/model_new/ from FastSpeech. We have removed the WaveGlow model from FastSpeech due to performance limitations.