Record first, transcribe after
Recording uses your browser's own recorder, so it works even on devices that are too slow for speech recognition. When you stop, you can play the recording back and save it as a compact audio file or as WAV. Transcription is a separate step: a Whisper speech model runs on this device and writes the text piece by piece, with a timestamp for each part.
Notes you can use
Download the transcript as plain text, as SRT subtitles with timings, or as Markdown notes with a heading, the date, the length and a timestamp every minute or so. Markdown notes open neatly in Obsidian, Notion, Bear and most note apps, and the timestamps help you find the part of the recording you need.
Private by design
Lectures, interviews and meetings often contain things that should stay in the room. Nothing here is uploaded. The recording lives in this tab until you save it, and the speech model downloads once and then works offline. Ask permission before you record other people, and follow your school's or workplace's rules.
Questions
How long can I record?
As long as your device has memory to spare, typically several hours. The recording is kept in this tab, so save it before you close the page. The screen is kept awake while you record, where the browser allows it.
Why is the browser asking for my microphone?
Recording needs your permission to use the microphone. The sound goes only to this page on your device.