01 / TWO ROUTES
There are two routes, and they need different things.
Extracting the text from a YouTube video sounds like one job. In practice there are two, and the difference is not cosmetic: one route reads words that already exist on the video, and the other listens to audio and writes the words down. The YouTube transcript downloader offers both, on the same page, with the link box at the top and a file upload underneath.
Which one is open to you depends on one question: does the video already have captions, and do you have the file? A public video with captions needs nothing but its address. A video with no captions needs the actual recording, which usually means it has to be yours. There is also a third path that catches the awkward cases, described at the end.
02 / THE LINK ROUTE
Route one: a link, when the captions exist.
Paste a watch link, a youtu.be link, a Shorts link, a live link, an embed link, or the eleven-character video ID. The tool asks YouTube for the caption tracks attached to that video, takes one, and shows the text as timed events, line by line.
What it does not do is listen. The link route never works from the soundtrack, so it cannot invent a transcript for a video that has no captions. When there is no track, the answer is a message saying so rather than an empty result — a distinction that matters, because an empty result looks like a bug while a message is an answer. If the video is your own, that message is also a signpost to the second route.
One premise of this route is easy to overlook: the link has to reach a video the server is allowed to see. When YouTube treats the request as automated and asks for a verification, the captions cannot be fetched, and the page says exactly that instead of failing silently.
03 / WHAT A TRACK IS
What a caption track is, in practice.
Two kinds of track exist on YouTube videos, and the difference shows in the text. An uploaded track is written by the creator. An automatic track is generated from the audio, and the page says its quality can vary with pronunciation, accents, dialects or background noise. Automatic tracks can also carry the marks of that process: repeated lines, a bracketed note where music plays, timings that run into each other.
You do not choose a language before you paste. When a video has an English track it is the one taken first, an uploaded track ahead of an auto-generated one; otherwise the first remaining track is used. For a single-language video that never comes up. For a video with several tracks it decides the language you read, and when a video has more than one track, the result shows a caption-language menu so you can switch among them.
And note the boundary of the product while you are here: the site is in English. The transcript itself is in the language the video uses; the interface, the articles and the way the pages are written are not translated.
04 / THE FILE ROUTE
Route two: upload the file when there are no captions.
This is the route for a video with no caption track, and it needs the recording rather than the address. The page takes a file in MP3, M4A, WAV, MP4, MOV or WebM, up to 20 minutes, and runs speech recognition on the sound inside it. You do not have to extract the audio first; the browser reads it out of the file.
Two premises go with it. The file has to be one you have, which in practice means your own recording or an export you were given: this page does not download a video from a link and then transcribe it. And the file has to sit inside the limits, because the size and length caps are enforced rather than suggested.
If you are weighing the routes before you start: a video with captions is a paste and a click, and a video without them is an upload. The second route needs no link, but it does need the file in hand.
The wider question of which route a given video needs is the subject of the article on transcribing a video with no captions, which compares the YouTube, TikTok and Instagram paths side by side.
05 / WHAT YOU GET
Both routes end in the same four options.
Whichever route filled the box, the result offers the same things: Copy for the clipboard, TXT for reading, SRT for a subtitle workflow, VTT for a web player. The TXT is the text on its own, with no time codes; SRT and VTT keep a number and a clock on every cue. Copy creates no file at all.
There is also the fallback path for the awkward case, and it is worth knowing before you need it. When YouTube refuses the caption request outright, the page offers a copy-from-YouTube route: open the video on YouTube, use its transcript panel, copy the text, paste it into the page. It is tidied in the browser, and a pasted transcript downloads the same way as a fetched one. That route is there for the moment the caption request is refused outright.
One thing none of these routes will do is edit. The tool does not change the words, it does not rearrange the timings, and it does not burn captions into a video. What it produces is text, which is what you asked for.
06 / QUESTIONS
Can I get text from a YouTube video that has no captions?
Not from the link. Upload your own copy of the file and the speech is transcribed from the audio.
Why does the tool not just listen to a YouTube link?
The link route reads caption tracks the video already has. Speech recognition runs when you upload a file instead.
Which language will the transcript be in?
An English track is taken when the video has one, otherwise the first remaining track. The site itself is in English.
What if YouTube refuses the request?
Open the video on YouTube, copy the transcript shown there, and paste it into the page. It is tidied in your browser and downloads like any other result.
READY WHEN YOU ARE
Start with the link or the file.
Paste the address, or upload the recording. Either way the words come back as text you can keep.
Get the text from a YouTube video ↗