← Transcript library

Transcript Generator

Turn speech in an audio or video file into an editable, timestamped draft. Your browser processes the recording on your device.

Up to 100 MiB per file. The first run loads about 55 MB of model and runtime data. Audio stays on this device.

Choose an audio or video recording

Start with a short passage where you know the words. MP3, WAV, M4A, MP4 and MOV can work when your browser can decode the audio inside. If a movie fails to open, export its audio as PCM WAV and select that file.

Check names and speaker turns

Play the recording beside the draft. Check names against your guest list, and mark the start of each change of speaker with a name or initials. The model produces timestamped passages. Speaker names are added by hand.

Save a corrected copy

Edit the draft in place, then download it as TXT. The first run downloads the speech model. Your recording and edited text stay in this tab’s memory until you clear the session or close the page. A downloaded TXT remains on your device.