Transcribe Audio

Transcribe turns one piece of audio — an uploaded file or a recording made on the spot — into a written transcript you can read and download. Nothing is stored; the result lives only on this screen.

At a glance

  • An audio file in eight formats, up to 25 MB
  • Or a microphone recording of up to 30 minutes
  • A timecoded, speaker-by-speaker transcript with an optional summary
  • Download as text or JSON — nothing is stored
Transcribe Audio

The Recording

The screen accepts one piece of audio at a time: a dropped file in any of eight formats up to 25 MB, or a microphone recording of up to 30 minutes with the elapsed time counting beside Stop Recording. Starting one input cancels the other, and Transcribe stays disabled while the microphone is open. If the browser refuses the microphone, the screen says exactly why — each kind of refusal has its own sentence.

The Transcript

The result reads as rows of timecode, speaker and text, each voice labelled Speaker 1, Speaker 2 and so on, with the summary above them when you asked for one. Download as text and Download as JSON hand it over whole, and an info line reminds you that nothing is kept.

  1. Provide the audioDrop an audio file, or press Record and speak; the screen holds exactly one recording, so starting either input replaces the other.
  2. Choose the optionsPick the language of the recording — or leave it on Detect automatically — and tick Write a summary as well if you want one.
  3. Transcribe and downloadPress Transcribe, read the result, and take it away as text or JSON before you leave the page.

Good to know — The transcript is never stored — a reload loses it, so download what you need before leaving.