Three words, often mixed up
People use captions, subtitles and transcripts almost interchangeably in conversation, and the meaning of the first two changes by region. In some countries, subtitles is the general word for timed on-screen text, and captions is a more specialized term. In others, the two are clearly separated. What matters is what the text contains and what it is for, so this article focuses on those points rather than on a single correct vocabulary.
Captions
Captions are timed text shown with the video that represent the spoken words and the other sounds needed to follow the program. They are designed for people who cannot hear the audio, so they commonly include:
- speaker identification when it is not clear who is talking
- descriptions of meaningful sounds, such as a phone ringing or applause
- music cues and tone indicators where they matter
Captions assume the viewer understands the language being spoken. Their main audience is deaf and hard of hearing viewers, although many other people use them, for example when they watch without sound.
Subtitles
Subtitles are timed text that present the dialogue, most often as a translation of speech in another language. The idea is that the viewer can hear the soundtrack, so there is less need to describe sounds. Subtitles are the usual choice when the audience hears the original language but does not understand it.
Subtitles in the same language as the audio exist too, sometimes called intralingual subtitles. When they follow the access conventions for deaf viewers, they are essentially captions under another name. When they only repeat dialogue, they may leave out sound information that some viewers need. Which one a viewer receives is often a matter of what the provider chose to include, so it can help to look at the label on the track.
Transcripts
A transcript is a written record of a recording, presented as a document rather than timed to the picture. It may include only speech, or speech plus sound and visual notes. Transcripts are read at the reader’s own pace and can be searched, copied, printed or opened by a screen reader or braille display.
Because a transcript is not tied to playback, it is useful in situations where timed text is not:
- a reader wants to scan a long recording quickly
- the person cannot watch or listen in real time
- the content is audio only, as with podcast transcripts for spoken shows
- someone needs to quote or study exact wording
A transcript does not show timing, so it cannot replace captions during playback. A viewer who needs to follow along in time cannot rely on a document alone.
A comparison at a glance
| Feature | Captions | Subtitles | Transcript |
|---|---|---|---|
| Timed to the video | Yes | Yes | No |
| Main purpose | Access for people who cannot hear | Understanding another language | Reading, searching and reference |
| Includes sound information | Usually | Usually not | Optional |
| Typical language | Same as audio | Often different | Same as audio, or translated |
These are typical patterns rather than fixed rules. A provider can decide to combine elements.
Choosing between them
A few simple questions help:
- Does the viewer need sound information? If so, choose captions that include it.
- Does the viewer understand the spoken language? If not, subtitles in a language they read are the usual answer.
- Will people want to read at their own pace? A transcript adds value, especially for long or technical recordings.
- Is a viewer unable to see the screen? Timed text will not help; consider audio description and a transcript they can read with assistive technology.
Many programs benefit from more than one. A lecture might have captions for deaf and hard of hearing students, subtitles in other languages for international learners, and a transcript for revision.
Related technical points
Timed text is delivered in several ways. Some arrive as separate files that the player loads, and others are part of the video. The distinction between the two affects what the viewer can adjust, and the choices are explained in the discussion of file formats and of how different subtitle tracks are offered. Timing for each line of text is a separate craft, since text has to appear and disappear in step with speech.
Another option sometimes confused with subtitles is a replacement voice track. Dubbing options are a different approach to a similar language problem, replacing the audio rather than adding text.
Quality matters for all three
Whichever form is used, accuracy counts. A caption that misses a key word or a transcript that confuses speakers can mislead a reader. Where the material is medical, legal or otherwise important, a qualified human reviewer should check it, especially when it was produced by automated tools.
Rules about which of these must be provided vary by place and by type of organization. This article offers general information only; consult the relevant authority or a qualified professional for what applies to you.