Working with YouTube
How to Get a Transcript from a YouTube Video
Most YouTube videos already carry captions, generated or uploaded, and the text can be pulled out as plain text or as timed subtitles. Here is what is available and what it is worth.
2 min read · Published 31 August 2026
There are two kinds of caption on YouTube and the difference decides how useful the text will be. One was written by a person and uploaded. The other was produced by speech recognition and has never been read by anyone.
Both come out as text. Only one of them can be trusted without checking.
Telling them apart
Open the caption menu on the video. An entry reading English is a supplied track. One reading English (auto-generated) is the machine's attempt. Where both exist, the supplied one is always the better source.
Auto-generated captions are usually accurate enough to follow and rarely accurate enough to quote. They have no punctuation to speak of, they invent plausible words where the audio was unclear, and they handle names and technical terms badly. Accented speech, two people talking at once, and music underneath all make it worse.
What you can take away
| Format | What it is | Use it for |
|---|---|---|
| Plain text | The words, no timings | Notes, quoting, translation, search |
| SRT | Text with start and end times | Subtitles on a re-edit, other players |
| VTT | The web subtitle format | Video on a web page |
Plain text is what most people actually want. The timings matter only if the text is going back onto a video.
A video with captions switched off has no text to retrieve. Nothing can be extracted that was never made, and where an uploader disabled captions entirely, the transcript does not exist on YouTube's side either.
Doing it
YouTube's own route is on the watch page: the three dots under the video, then Show transcript. It appears in a panel beside the video and can be copied out, though it arrives with a timestamp before every line, which is tedious to strip by hand for anything long.
The tool here takes the link and returns the text already cleaned up, or the timed SRT if that is what you need. It reads the same captions YouTube would show you, so what it cannot get is what was never published.
A use worth mentioning
Transcripts of your own videos are worth keeping for reasons that have nothing to do with subtitles. They make an archive searchable, they turn a video into the first draft of a written post, and they let you find the moment you said something without scrubbing through an hour of footage.
Somebody else's transcript is their words. Quoting with attribution is ordinary practice; republishing the lot as your own article is not, and a machine transcription does not change whose words they are.
Common questions
Can I get a transcript of any YouTube video?
Only where captions exist. Most videos have at least auto-generated ones, but an uploader can disable captions entirely, and where they have, there is no text to retrieve on any route. Live streams that have not finished processing often have none yet either.
How accurate are auto-generated captions?
Good enough to follow, not good enough to quote. There is little punctuation, unclear audio becomes a plausible guess rather than a gap, and names and technical terms are frequently wrong. Anything being published needs reading against the audio first.
What is the difference between SRT and plain text?
SRT carries start and end times for every line so a player can show each one at the right moment. Plain text is just the words. If the text is going back onto a video, you want SRT; for notes, quoting or translating, plain text is easier to work with.
Does a transcript include the speaker names?
Not from auto-generated captions, which produce one continuous stream with no idea who is talking. A caption file written by a person may label speakers, and if it does, the labels come through with the rest of the text.