Transcript vs Subtitles
Transcripts, subtitles, captions and closed captions are four different things. Here is what each one is for and which you actually need.
The Scriptorium team ·
A transcript is a document you read. Subtitles are timed text displayed over a video. Captions are subtitles that also describe non-speech sound. Closed captions are captions the viewer can switch off. They overlap, and choosing the wrong one means doing the work twice.
The four things
| What it is | Where it appears | |
|---|---|---|
| Transcript | A readable document of everything said | A page, a document, a file |
| Subtitles | Short timed chunks of dialogue | Over the video |
| Captions | Subtitles plus non-speech sound and speaker IDs | Over the video |
| Closed captions | Captions the viewer can turn on or off | Over the video |
The distinction between subtitles and captions is the one people miss. Subtitles assume you can hear the video and just need the words — historically, for translation. Captions assume you cannot hear it at all, so they include [door slams], [laughter] and speaker labels. For accessibility, captions are the requirement, not subtitles.
Which one do you need?
For accessibility
Captions, on the video. Ideally a transcript as well: some people prefer reading the whole thing to following timed text, and screen readers handle a transcript far better than a caption track.
For search engines
A transcript, published as text on the page. Caption files are not reliably indexed as page content, and a transcript gives a search engine full sentences in context rather than fragments.
For social video
Captions, usually burned into the picture. Most feed video plays muted by default, so text that only appears when someone enables it will not be seen.
For quoting or referencing
A transcript with timestamps. Subtitle files are chopped into display-sized chunks that break mid-sentence, which makes them awkward to read and worse to quote from.
For translation
Subtitles, generated from a translated transcript. Translate the transcript, then time it, rather than translating fragments.
How they relate
Subtitles are made from a transcript. The transcript is the source document; producing subtitles means splitting it into chunks short enough to read on screen and giving each chunk a start and end time.
Which is why the efficient order is: transcribe once, correct once, then export both. Correcting a subtitle file line by line after the fact is much slower than correcting a transcript and regenerating.
The file formats
- SRT. The most widely supported subtitle format. Plain text, numbered blocks with start and end times. Accepted almost everywhere.
- WebVTT (.vtt). The web standard, and what HTML5 video uses. Supports speaker voice tags and styling. YouTube accepts it.
- Plain text. A transcript with no timing at all.
- Timestamped text. A transcript with a time against each paragraph. For reading and citing, not for display over video.
If you need both, export the transcript for the page and SRT or WebVTT for the player. Same source, two outputs, no duplicated work.
A note on burning captions in
Burned-in (or "open") captions are rendered permanently into the picture. Viewers cannot turn them off, which is the point on social feeds and the problem everywhere else — they cannot be translated, resized, or disabled by someone who finds them distracting. Use them for feed video and use a caption track for everything else.
Scriptorium exports subtitle files; it does not burn captions into video. That step happens in a video editor or on the platform you upload to.