Extract hardcoded subtitles
from any video
Burned-in subtitles in, editable SRT out — read on your own device. No uploads. No server.
- No upload — videos never leave your device
- No sign-up — no account, no limits
- No tracking — no analytics, no beacons
- 18 tools — extract, edit, convert, clean
This site has no server and no upload endpoint — open it and use it; close the page and nothing is left. Tool search: ⌘K
Everything you can do here
One job done properly — getting subtitles out of video — and a full set of tools for everything that comes after.
How it works
From a burned-in video to an editable subtitle file in four steps.
Drag in any video with burned-in subtitles — MP4, MOV, MKV, WebM and the rest. It stays on your device; the page reads the frames directly.
Drag a box around where the text sits on the picture. A tighter box means fewer things to misread — the default already lands on the usual spot.
The video is scanned for changes and only the frames that matter are read — so a whole episode takes seconds, not hours. Watch it happen, or stop it any time.
Clean up a line in place, then take your subtitles as SRT, VTT, ASS, plain text — or send them straight to the editor to time and polish.
Questions people ask
Short answers, no fine print.
Is my video uploaded anywhere?
No. The video is opened by the page itself and every frame is read on your device — there is no upload step and no server to upload to. Turn off your internet after the page has loaded and extraction still runs. The one download it makes is the recognition model (about 30 MB), fetched once and then kept offline by your browser.
What kinds of subtitles can it read?
Anything that is permanently part of the picture: hardsubs, open captions, burned-in translations, lower-thirds — the kind no player can switch off. It also picks up on-screen text that behaves like a caption (signs, callouts) if you box that area. Subtitles stored as a separate track inside the file or as a sidecar .srt are already text — use the editor or the converters for those.
How accurate is it?
On clear video with ordinary fonts, recognised lines regularly come back in the 90s. Quality drops with blur, heavy compression, unusual fonts and busy backgrounds — boxing the subtitle area tightly is the biggest single improvement. Every line comes with its confidence score so you can see what to double-check.
Which languages does it work with?
Over 60 — Chinese (Simplified and Traditional), English, Japanese, Korean, French, German, Spanish, Russian, Arabic, Hindi and many more. Pick the language before extracting; it only affects recognition, not timings.
How long a video can I throw at it?
As long as your browser can play. Extraction takes roughly 1–2 seconds per minute of video on a laptop — most of the time goes on reading the frames where the subtitles actually change. You can queue up to eight videos and take all the results as one ZIP.
Can I edit the subtitles afterwards?
Yes — click straight into a line to retype it, fix a timecode, or delete it, right next to the video. For timing work (shifting, splitting, merging cues, find-and-replace) send the result to the built-in editor; it opens over the video with the subtitles drawn where they will play.
Got a video with burned-in subtitles?
Turn it into an editable subtitle file in about the time it takes to make coffee.
Extract subtitles — free