How to set up a custom Speech Engine

The easiest way to create a custom Speech Engine for Izabela is to create a local server that will provide the following endpoints:

• /list-voices in order to list available voices
• /synthesize-speech in order to synthesize audio from text

This guide will show how to run a simple Node.js server written in JavaScript which provides these endpoints.

Prerequisites

• Node.js
• Command Line Interface (Terminal)

How to run the example sample

After downloading Node.js your Terminal should have access to new commands including:

• node in order to run JavaScript in a terminal
• npm in order to install JavaScript dependencies

For now, create a new folder containing the files package.json and index.json from the example sample found here.

Right click on the folder and click on "Open in Terminal".

This should open the default Windows Terminal in the current directory.

In this terminal you should be able to run these commands:

• npm install in order to install the depencies from our package.json
• npm run start in order to run the start script from our package.json

You can check if the custom Speech Engine works by pasting the API Endpoint in Izabela:

Adding custom Speech Engines

You can add any Text-to-Speech solution as long as your local server returns the correct data for Izabela.

You can find how requests and responses should look like on the Custom Engine API example sample page.

You can use this custom ChatGPT prompt if you don't know where to start: https://chatgpt.com/share/67028671-67e4-800d-8a76-d365a62758ad