This page is for podcasters and for anyone who wants an episode as text: a transcript to publish next to the episode, a base for show notes, quotes for social posts, or subtitles for the video version. Text also makes an episode searchable: people can find what was said without listening to the whole hour.
Drop the episode file here or tap Choose a file. Audio in MP3, M4A, WAV, OGG, OPUS, FLAC, AAC or AMR works, and so do video episodes in MP4, MOV, WebM, MKV, AVI or 3GP. An episode can run up to 120 minutes and the file can be up to 2 GB; from a big video your browser can take out the sound so the upload is smaller. Links to Spotify, Apple Podcasts or YouTube are not accepted: you need the file itself.
You get the text split into phrases, each with its timestamp, so you can find the minute where a topic starts. Download TXT for your website, Word (DOCX) for editing, or SRT and VTT for the video version. The language is detected on its own, and the whole transcript can be translated into 45 languages.
For an episode of 10 minutes or more, “Key points” appears above the text: the key ideas, the lines said word for word and the numbers and dates mentioned. It is a starting point for show notes, not a finished text, so read it and keep your own voice. You can also ask in your own words, for example for five short quotes for social posts.
If you release often, pick up to 20 episodes at once. Each is transcribed separately, and when they are ready you see the text of each, tick the ones you need, download them together or merge them into one text. A share link sends the transcript to a co-host or an editor without the audio file and works for 30 days.
Two voices close to their microphones come out best. Hosts talking over each other, music under speech and a guest on a weak phone line cause mistakes, so check names, brands and numbers before you publish. The file is deleted from our server after 24 hours and the text stays in the library of this browser. It is free and needs no sign-up; to keep bots out, one connection can run up to 60 AI tasks an hour.
