whisper-turbo

Why we included this project

Whisper Turbo runs OpenAI's Whisper directly in the browser or inside an Electron app, doing all transcription on-device with WebGPU acceleration. Because nothing is sent to a server, it fits workflows where privacy matters: meeting notes, patient intake calls, anything with strict data-handling rules. The underlying Rust library, Ratchet, keeps it light compared to Python-based Whisper setups, and the npm package drops into a web project without much fuss. Just note that WebGPU is officially supported only in Chromium-based browsers on Windows and macOS, so it's worth testing on the platforms you actually ship to.