Audio model · from 1 credit

Transcribe audio or video to text with ElevenLabs Scribe v2

Scribe v2 is ElevenLabs' transcription model: it turns a recording, an interview, a podcast or a video's soundtrack into text, in more than 90 languages. Every word is timestamped, which gives a ready-to-use .srt subtitle file, and voices are told apart: in a conversation, each turn comes out on its own numbered line. Notable sounds, a laugh for instance, are noted in brackets. On the canvas it lives in the Transcribe mode of the Audio node: wire in a generated voice, a video or an imported file, and the editable text it returns flows like a Text node into a prompt or a language model. On a short clip, set the language rather than letting it guess.