Open an audio or video file in the web app and choose Transcript at the top of the preview. The panel says how long the recording is, what the transcript costs and how many credits you have left. Choose Transcribe.
- Cost: one AI credit for every started minute. Nothing is charged if it fails.
- Language: leave Detect it, or name the language that is spoken. Naming it helps with accents and short recordings.
- While it works you can close the window. A notification tells you when the transcript is ready.
Reading it
Click a line and the player jumps to where it is said; the line being played is marked. Type in Search in the transcript to keep only the lines with those words. The buttons at the top of the panel copy the text or download it as text (.txt), captions (.vtt) or subtitles (.srt). A video shows its transcript as captions in the player.
Once a recording has a transcript, Ask AI answers questions about it, and search by meaning finds what was said in it.
Good to know
- Speech recognition makes mistakes, more of them with names, figures, accents and poor sound. For what matters, listen to the recording.
- Only the sound is sent to the speech model, and only when you ask.
- Everyone who can open the file can read its transcript. Anyone who can open the file can ask for one, and pays for it with their own credits.
- Transcribe again (under …) replaces the transcript, for another language for instance, and costs again. Remove the transcript deletes the text and keeps the recording.
- After you upload a new version of the file, the transcript is marked as made from earlier content. Choose Transcribe what the file holds now.
- Very long recordings are not transcribed; the panel says what the longest is. A recording file can be 4 GB at most.
- Transcripts are in the web app. The phone and desktop apps do not show them yet.
If the preview has no Transcript button, the service has no speech model set up.