On-device Automatic Speech Recognition

About

A ready-to-use, minimal app that converts any speech into text, processed entirely locally using Ratchet. You can deploy it as a website or build it as a desktop application. Try the web version here.

Important: This application requires access to the WebGPU API. If you're using it on the web, please ensure your browser supports WebGPU.

Getting Started

Prerequisites

• Node.js
• Npm

Run it locally

• Clone the project.

    git clone https://github.com/Hugo-Dz/on-device-transcription.git
    cd on-device-transcription

• Install the dependencies.

    npm install

• Start the dev server

    npm run dev

Build the desktop application

• Build the Svelte application.

    npm run build:svelte

• (optional) Start the Electron application in dev mode.

    npm run dev:electron

• Build the Electron executable.

    npm run build:electron

Resources

• OpenAI Whisper
• Ratchet SDK
• Electron with SvelteKit

License

MIT License Hugo Duprez