Whisper-web
Browser-based speech recognition tool that runs OpenAI Whisper entirely in the browser with no server required.
About
Whisper-web brings OpenAI's Whisper speech recognition model directly into the browser using the Transformers.js library, so audio is transcribed on your device without sending data to a server. You can try it immediately through a hosted demo on Hugging Face Spaces or self-host it as a web app built with TypeScript and Vite. Experimental WebGPU support is available on a separate branch for faster in-browser processing on compatible hardware. The project is free and open source, making it useful for developers who want client-side, privacy-respecting speech-to-text without any backend infrastructure.