Transcribe Checking device… Preparing…

The first run fetches the model once, then it stays cached on this device.

Leave the tab open and it keeps working in the background.

Transcribe meeting audio in your browser.

Pick an audio or video recording and read it back as text. The speech model runs on this device, so the file is never uploaded.

Pick a file. Read it back as text.

The whole run happens inside this tab. There is no queue on a server, no account to make, and nothing to install.

  1. Drop a recording onto the panel above, or click it and choose one. Audio or video, either way.
  2. The speech model loads once, about 75 MB, and the browser keeps it. Every run after that starts straight away.
  3. Watch the text arrive while it works, then copy the transcript or save it as Markdown with the timestamps in place.

Your audio stays on your Mac.

The file is decoded and read here, and it is never sent anywhere. You do not have to take that on trust.

Open the network tab in developer tools, run a transcription and watch for outgoing requests. After the one-time model download there are none, and pulling the network cable out mid-run changes nothing about the result.

  • No uploadThe recording stays inside this browser tab.
  • No accountAnd no tracking on this page.
  • Works offlineOnce the model is cached.

The things people ask first.

Does my file get uploaded?

No. It is read inside this browser tab and it is never sent anywhere. Once the model has been fetched, the page makes no further requests, which you can watch for yourself in developer tools.

Which files can I use?

Audio and video. mp3, m4a, wav, mp4, mov, ogg, flac and webm all work, and the audio track is taken out of a video file for you.

What happens on the first run?

The page fetches the speech model once, around 75 MB, and the browser keeps it cached. Every run after that starts straight away, and the tool keeps working with no connection.

How long does a recording take?

That depends on your machine, because the work happens on your own processor instead of on a server. A short clip comes back quickly and a long meeting takes a while, and the text arrives as it goes, so you can read along. Leave the tab open and it keeps working in the background.

Is there a sign-up or a file limit?

Neither. There is no account, no email and no cap on this page. The only ceiling is your own machine and the memory your browser hands to the tab.

How is this different from the Aside app?

This page reads a file you already have. Aside records the meeting itself on your Mac, tells the speakers apart and writes the note into a folder you choose, as Markdown.

Aside does this while the meeting happens.

The same on-device transcription, running on your Mac as the call goes on, with every note saved as Markdown in a folder you choose.

Requires macOS 26 or later. Apple silicon, or Intel with much slower transcription.