This project provides a minimal, ready-to-use application for on-device speech-to-text transcription, targeting users who need local processing for privacy or offline capabilities. It leverages the Ratchet SDK and can be deployed as a website or a desktop application.

How It Works

The application processes speech entirely locally using the Ratchet SDK, which is built upon OpenAI's Whisper model. This approach ensures data privacy and offline functionality, eliminating the need for cloud-based APIs. The architecture is designed for minimal resource consumption and ease of deployment.

Quick Start & Requirements

• Primary install/run command: npm install followed by npm run dev for the web version.
• Prerequisites: Node.js, Npm.
• WebGPU API access is required for the web version; ensure browser support.
• Building the desktop application involves npm run build:svelte and optionally npm run dev:electron or npm run build:electron.
• Official web version: https://on-device-transcription.vercel.app/

Highlighted Details

• On-device processing via Ratchet SDK (Whisper).
• Deployable as a website or desktop application (Electron).
• Minimal and ready-to-use design.

Maintenance & Community

• Developed by Hugo Duprez.
• No specific community channels or roadmap links are provided in the README.

Licensing & Compatibility

• License: MIT License.
• Compatible with commercial use and closed-source linking due to the permissive MIT license.

Limitations & Caveats

The application requires WebGPU API support, which may limit browser compatibility. The README does not detail performance benchmarks or specific hardware requirements beyond browser support.