Extract the caption track, not a new recording
Nearly every public video on YouTube carries a caption track: the spoken words stored as text, with the moment each line appears. The player shows that track a line at a time and never gives you the whole thing. This extractor asks for the track directly and puts all of it on one page, in the original language of the video, in the order it was spoken.
That is a narrower job than transcribing. No audio is processed and no wording is changed. What comes out is the caption data YouTube already holds, laid out so you can read it from start to finish instead of watching for it.
Written captions and automatic ones
YouTube subtitles come from two places. Some are written or corrected by the creator, and those are usually accurate, including names and technical terms. The rest are produced automatically by speech recognition, and their quality follows the audio: clear speech in a quiet room extracts cleanly, while heavy accents, music or several people talking at once leave errors in the text.
The extractor takes whichever track is available and prefers the creator version when there is one. Both kinds carry timings, so both arrive with timestamps.
Extracting YouTube subtitles in the browser
The usual way to extract subtitles from YouTube is a command line tool that you install, configure and then run per video. That works, but it is a setup step before every first use, and it produces a file you still have to open to see what is inside. Here the order is reversed: the text appears first, you read it, and only then decide whether it is worth keeping.
Because it is a web page, the same process works on a phone, a work laptop where you cannot install software, or a shared machine. Regular videos, Shorts and livestream replays are all handled the same way.
Free to read, paid to take away
Extracting and reading costs nothing. Without an account the limit is 2 videos a day, and video length is not restricted, so a three hour recording counts the same as a two minute clip. Taking the text off the page is where a plan applies: copying to the clipboard and downloading the extracted track as TXT, SRT or VTT are included in Basic at 6.99 US dollars a month and Pro at 14.99. The pricing page lists the rest of what each plan covers.
What the extracted text is used for
Editors pull the captions of their own uploads instead of retyping them into a timeline. Translators start from the source track before producing another language. Students keep lecture text they can search rather than notes they have to rewatch. Writers quote a speaker from the record instead of from memory, and the timestamp next to each line points straight back to the passage in the video.
The file format follows the purpose. TXT gives plain prose for a document. SRT and VTT keep the timings intact, which is what a video editor or a web player expects.
Shorts extract like any other video
A YouTube Short is not a separate kind of upload as far as captions are concerned. It is a vertical video under a minute, it gets the same automatic caption track as everything else, and a /shorts/ link extracts exactly like a /watch link – same speed, same timestamps, same free allowance. The one thing worth knowing is that a Short's captions are auto-generated far more often than a long upload's, because few creators go back and correct sixty seconds of text.
Short-form video from elsewhere is a different problem, not a smaller one. TikTok and Instagram Reels publish no caption track for anyone to read, so there the audio itself has to be transcribed. Those have pages of their own, and their own free allowance.
Beyond the raw caption text
Each of your 2 free videos a day comes with an AI summary, and paid plans add more AI work on top of the extracted transcript. You can ask for a summary whose key points each carry a timestamp, put questions to the video, or search for the moment a particular subject comes up. The batch page processes a whole list of links in one run – 10 on Basic, 30 on Pro, 50 on Max – which is the practical option when you are working through a series rather than a single video.