What is a video transcript? Meaning, transcript vs captions, and why editors use one

A video transcript is the spoken content of a video written out as text, usually with timestamps that tie each line back to the moment it was said. Captions are different: they are short pieces of that text shown on the picture, timed to the speech, in the same language. Subtitles, in the strict sense, are captions translated into another language. A transcript lives next to the video or underneath it, reads like a document, and can be searched, skimmed and, with the right tool, edited. That last point is why people who cut talking-head videos care about transcripts at all: when every word has a time, deleting a word can delete the video behind it.

The three words, and where they overlap

The terms get mixed up because they share the same raw material. The W3C, which writes the web accessibility guidelines, draws the lines like this (W3C WAI): captions are text “for the same language as the spoken audio”, and subtitles are “for spoken audio translated into another language”. It also notes that some regions use “subtitles” for both, which is why a British viewer and an American viewer can mean different things by the same word.

A transcript is the third thing. The W3C defines a basic transcript as “a text version of the speech and non-speech audio information needed to understand the content” (W3C WAI). A descriptive transcript goes further and also describes what is on screen, for people who are both deaf and blind.

	Transcript	Captions	Subtitles
Where it appears	Beside or below the video, or in a separate file	On the picture, in sync with speech	On the picture, in sync with speech
Language	Same as the audio	Same as the audio	Translated
Form	Continuous text, often with timestamps	Short timed chunks	Short timed chunks
Read without the video	Yes	No	No
Used for editing	Yes, in text-based editors	Rarely	Rarely

One practical consequence: a caption file is a transcript cut into pieces with timings. If you have captions, you have a transcript. YouTube works exactly this way. Its “Show transcript” button is available for “any video that has captions”, and clicking a line of the transcript jumps the video to that moment (YouTube Help).

Why editors want the transcript, not just the captions

Captions help the viewer. The transcript helps you before the viewer ever sees anything.

A ten-minute take is hard to scan on a timeline because the waveform all looks the same. The same take as text is two pages you can read in a minute. You can see the false start in the second paragraph, the sentence you said three times, and the tangent that runs from 04:12 to 05:40. Finding those on a timeline means scrubbing and listening. Finding them in a transcript means reading.

Text-based editing takes this one step further. Adobe announced the feature for Premiere Pro in February 2023, describing it as transcribing footage on import so you can “cut and paste sentences” and have clips in the timeline follow your edits (Adobe). Descript describes the same idea: “delete, copy, and paste in the transcript to edit your audio and video” (Descript). On a phone, slicecut does it for talking-head takes: import a video from the gallery, the words come back a few seconds later, and tapping a word strikes it from the video, with the cut landing right before the next word.

The reason this suits talking-head videos in particular is that almost every cut in one is a speech cut. You remove a repeated sentence, a filler word, a long pause. Those are all things you can see in text. For a silent action shot, where the cut depends on movement, a timeline is still the right tool. Our guide to editing a talking-head video walks through the passes in order, and the comparison of CapCut vs Descript sets the two approaches side by side.

How to get a video transcript

You have three realistic routes. Which one fits depends on whether you want to read the transcript, publish it, or edit with it.

1. Let the platform make it. YouTube generates automatic captions with machine learning, and says plainly that “the quality of the captions may vary” and that you should “always review automatic captions and edit any parts that haven’t been properly transcribed” (YouTube Help). Once the captions exist, the transcript view exists too. This is free and takes no effort, but it only works after upload, so it does not help with the edit.
2. Run a speech recognition model yourself. OpenAI’s Whisper is “a general-purpose speech recognition model”, and its code and model weights are released under the MIT licence (OpenAI). It is free, it is not a phone app, and it takes some setting up. It suits people who want a text file and are comfortable with that.
3. Use an editor that transcribes. Descript says its transcription “delivers up to 95% accuracy” (Descript). Premiere Pro transcribes on import (Adobe). slicecut sends only the audio of a take off for transcription and returns the words into the project. In all three, the transcript is the editing surface, not a by-product.

Upside: Automatic transcripts take seconds, cost little or nothing, and are good enough to navigate by.

Downside: Names, jargon and crosstalk get mangled, and a transcript you publish still needs a read-through.

How to write a video transcript by hand

Sometimes you need a transcript for a page, a blog post or an accessibility requirement, and the automatic one is not clean enough. Writing one from scratch is slow, so the usual method is to generate first and correct second.

1. Get an automatic transcript by any route above, so you are correcting rather than typing.
2. Play the video at normal speed with the text beside it and fix every wrong word. Do not trust the proper nouns.
3. Decide how verbatim you want to be. A clean transcript drops “um” and false starts; a verbatim one keeps them. For a published transcript, clean is nearly always right. For an edit, you want the ums left in so you can strike them.
4. Add speaker names if more than one person talks, and timestamps every paragraph or so, so a reader can find the spot in the video.
5. If you need captions as well, keep the timed version separate. Captions have line length limits that a readable transcript does not.

The long version of this, including what to clean and what to leave, is in our guide to writing a video transcript.

What a transcript shows you about your own video

Read your own take as text once and you will see patterns the timeline hides. The same opening phrase at the start of every paragraph. A filler word that appears every third sentence. A good sentence buried after two bad attempts at it.

Those are edit decisions, and the transcript lets you make them in order. Strike the repeats first, since repetition is what viewers notice. Then the filler words, and only the ones that do nothing: our piece on why filler words are not the enemy makes the case for keeping some. Then shorten the pauses. Each of those steps has its own guide, including how to remove filler words from a video and how to make jump cuts from a transcript.

Is a transcript required?

For most creators, no. The web accessibility guidelines require captions for prerecorded video at Level A, the baseline level (W3C WCAG 2.2), and organisations that have to meet those guidelines usually provide transcripts as well. For a Reel or a Short, nobody will ask you for one. But you will probably want one anyway, because it is the fastest way to see what you actually said, and the fastest way to cut it.