Transcription from audio, locally
A speech recognition model downloaded once and kept in cache. The panel tells you how much it downloads and how long it takes before starting, instead of leaving you in front of a spinner.
This translation is a draft not yet reviewed by a person: it may contain mistakes.
Subtitles for your videos
Written by hand, pasted from a script, or transcribed from audio. Transcription runs on your computer: the audio is never uploaded to any server, ours or anyone else's — not because we promise it, but because it never leaves the browser.
Open the Video Studio
Opens in the browser, no account.
Che cosa fa
The subtitle follows the voice
Every line has a from and a to in seconds, editable by hand. «Go» takes the video exactly there, so you check the sync by hearing it instead of calculating it.
From a script you already have
If the text is already written, paste it and «Split onto the video» spreads it over the duration. For prepared speech it's faster and more accurate than any automatic transcription.
SRT and VTT, the formats everyone opens
Import subtitles made elsewhere and export in the two formats YouTube, players and social networks accept. No format of ours you can't get out of.
Or burned into the video
Written into the picture: they show even where no player loads a subtitle file — that is, on almost every social network. And if burning fails, the video still exports without them, and it says so.
It warns you when they're unreadable
Lines too long, overlapping timings, a subtitle that lasts a third of a second. These are the mistakes you don't notice while writing and only notice watching the finished video.
I limiti, scritti qui
Dichiarati prima, non scoperti dopo.
- Automatic transcription gets confused by background noise, strong accents and overlapping voices: subtitles need a read-through before publishing. Saying it now is more useful than finding out later.
- The model is the smallest of its family — 42 MB, downloaded once. That choice is what makes transcription possible in everyone's browser; on hard audio a bigger model on somebody else's server would do better — and in exchange it would have your voice.
- It runs on your computer: on a slow machine or a phone from a few years ago, transcribing a long video takes patience. That's the price of not uploading the audio anywhere.
- It doesn't translate: subtitles come out in the language being spoken.
- The Video Studio works in the browser and doesn't save your footage on our servers: closing the page loses the edit. Export before you leave.