Hi Product Hunt! I built Whisper Web for people who need useful transcription without sending the source recording to another service. The media is decoded and transcribed in your browser. The first run downloads a Whisper model, and later runs can reuse the browser cache. You can choose Tiny, Base or Small, use WebAssembly or WebGPU where supported, edit the transcript, and export TXT, JSON, SRT or VTT. There is also a dedicated local workflow for files up to 1 GB or one hour, plus an MP4-to-MP3 tool. A few honest boundaries: the app still needs the network to load site assets and model files; remote URLs must allow browser access; and live meeting bots, speaker labels and cloud collaboration are not part of the product. I would especially value feedback on setup clarity, browser compatibility and where the local-processing model feels most useful.

Whisper Web

Private speech-to-text that runs in your browser

Private speech-to-text that runs in your browser