Definition: Captions are synchronised text for everything you can hear in a video: the speech, who is speaking, and the sounds that matter for understanding it. WCAG requires them for prerecorded video in 1.2.2 at Level A and for live broadcasts in 1.2.4 at Level AA [1][2].
German: Untertitel · Also: closed captions, subtitles for the deaf and hard of hearing (SDH) · Last reviewed: 27 September 2026
Captions are not the same as subtitles
In German both are called Untertitel, and in British English "subtitles" is often used for both. The distinction still matters. Subtitles translate speech for people who can hear the audio but do not understand the language. Captions are for people who cannot hear the audio, so they also include cues such as [doorbell] or [quiet music] and name the speaker when that person is off screen [3].
There are also two forms. Closed captions are a separate text track alongside the video; viewers switch them on and can often change their size and colour in the player. Open captions are burned into the picture: always visible, but impossible to enlarge or to read out as text.
Who they matter to
Deaf and hard of hearing people need captions to follow a video with sound at all. For many Deaf people in Germany, written German is a second language after German Sign Language, which makes clear, readable captions all the more important. Captions also help anyone watching without sound, on a train for example, and people learning a language.
What good captions have
- Complete: everything that is said, not a summary.
- Synchronised: text and sound line up in time.
- Speaker changes marked whenever it is not obvious who is talking.
- Meaningful sounds described, such as an alarm or laughter.
- Corrected: automatic captions tend to get names, numbers and technical terms wrong.
In code
On the web, captions usually live in a WebVTT file attached to a <video> with the <track> element [4].
<video controls>
<source src="film.mp4" type="video/mp4">
<track kind="captions" src="film.en.vtt"
srclang="en" label="English" default>
</video>
WEBVTT
00:00:01.000 --> 00:00:03.500
[doorbell]
00:00:04.000 --> 00:00:07.000
ANNA: Are you coming to the meeting?
Common mistakes
- Uncorrected automatic captions. They look finished and rarely are.
- Captions only on the recording. A live stream needs them while it is live.
- Captions instead of audio description. Captions serve people who cannot hear; blind people need audio description.
- Burned-in text only, in tiny type with weak contrast.
How Reviseberg handles it
Reviseberg does not watch videos, so whether captions are right – complete, accurate, in sync – is not something it decides. What it checks on every crawled page: whether a video of the page's own with a sound track offers a captions track at all, and whether that track loads and holds captions; for a YouTube or Vimeo embed, what the platform's public information says – no track at all is a failure, automatic captions only or information it cannot read are a question; and for a live stream, it asks, and reports a failure where the stream's own manifest lists no captions. Anything it cannot decide is listed as a potential issue and never counted as a pass. Criteria 1.2.2 and 1.2.4 therefore count as only partly testable by automation: they stay "partial" until a person records a result with its method, such as "watched every video with sound off and captions on". Guided manual tests set out the procedure step by step, and a recorded result is valid for twelve months.
Related terms
Audio description · Transcript · German Sign Language · Plain language · Manual accessibility testing