How to extract subtitles from a video
Two very different jobs share the same name. Work out which one you have first.
"Extract the subtitles" means one of two things depending on your file, and the tools for them have nothing in common. If the subtitles are already inside the file as a separate track, extraction is a copy operation that takes a second. If the subtitles only exist as speech — or as pixels burned into the picture — nothing can be extracted, and what you actually need is to create a subtitle track from scratch.
Open the subtitle editor → Creates a subtitle track from the speech in your video, which is what to do when there is no track to extract.- 100% freeNo account, no minute quota, no paid tier.
- Nothing is uploadedYour video is read on your device and never reaches a server.
- No catch on the outputNo watermark, no length cap, no locked export formats.
- Updated in the openEvery fix is written down on the changelog page.
Do you have a soft subtitle track?
Soft subtitles are a separate stream inside the container, alongside video and audio. MKV files very often have them; MP4 files sometimes do. In a desktop player like VLC or mpv, you can tell in a second: if the subtitle menu lists named tracks you can switch between and turn off, they are soft, and they can be extracted losslessly.
Extracting them is a job for a desktop tool — MKVToolNix for Matroska, or ffmpeg with a stream copy. Both take seconds and neither re-encodes anything. This site does not do that extraction: it is a five-second command-line job that desktop tools already do well, and building a worse browser version of it would be adding a feature rather than solving a problem.
If the subtitles are burned into the picture
Burned-in (hardcoded) subtitles are pixels. There is no text to extract — you would have to run optical character recognition on every frame, which produces a mess of duplicates and misreads and is genuinely hard to do well.
In almost every case where somebody wants the text from a video with burned-in subtitles, transcribing the audio is faster, more accurate and easier to correct than OCR would be. That is what the editor here does.
If there is no subtitle track at all
This is the usual situation, and it is not extraction, it is creation. Drop the video into the editor, let speech recognition run through the audio, and you get a timed subtitle track: text with a start and an end on each line. It takes a few minutes on a laptop and none of it leaves your machine.
The result is not a transcript of what a human would have typed — it is a first draft that gets most of the words right and needs a pass for names and punctuation. That pass is where the waveform timeline earns its place, and it is far less work than either OCR or typing from scratch.
Export as SRT if the file is going to a platform or a player, VTT if it is going into a web page, or plain text if what you actually wanted all along was the transcript rather than the timings.
Frequently asked questions
Can this site pull an existing subtitle track out of an MKV?
No. Use MKVToolNix or ffmpeg on the desktop — both copy the track out losslessly in seconds. This site creates subtitles from speech, which is the other job.
Can it read subtitles burned into the picture?
No, and OCR on video frames is rarely the right approach anyway. Transcribing the audio gives a better result with less cleanup.
How do I tell if my file has a soft subtitle track?
Open it in VLC or mpv and look at the subtitle menu. Named, switchable tracks are soft. If the only option is Disable and text still shows on the picture, they are burned in.
Does the video need to be a particular format?
MP4, MOV, WebM and MKV all work, along with most audio files. If the browser cannot decode the audio codec, the editor says so rather than failing silently.