Text, SRT or VTT
Download the transcript as plain text, in paragraphs, to read, quote or paste into notes. Or download it as subtitles: SRT for YouTube, Premiere Pro, DaVinci Resolve and most players, VTT for web players and Vimeo. Subtitles are named after your video, so a player finds them beside it.
Every line timed, every word fixable
The transcript comes in short lines, each with its start and end. Click a line to jump to it, correct a name or a word, change its times or delete it, and the downloads follow. As the video plays, its line shows over it, and in languages written with spaces the word being spoken lights up.
English and dozens of other languages
The language is recognized from the first words spoken. English is transcribed by NVIDIA's Parakeet, one of the most accurate speech recognition models, and on a graphics card many times faster than the video plays. Spanish, French, German, Italian, Portuguese, Japanese, Chinese and dozens of other languages are transcribed by OpenAI's Whisper.
Private by design: it runs on your computer
The speech models run in your browser, on your graphics card through WebGPU, or on your processor where there is none. Nothing is sent to a server: your file is read from your disk and the transcript stays with you. The first transcript downloads the models once; your browser keeps them for the next one.