Definition: A transcript is a full text version of audio or video content: everything that is said, who says it, and the sounds that matter. For audio-only content such as podcasts, it is the usual alternative under WCAG 1.2.1 at Level A [1]. A descriptive transcript also covers the important visual content of a video.
German: Transkript · Also: descriptive transcript · Last reviewed: 27 September 2026
Basic and descriptive transcripts
The W3C distinguishes two kinds [3]:
- Basic transcript: the text of all speech and meaningful sound, with speakers named. This suits podcasts and other audio-only content [4].
- Descriptive transcript: adds a description of what happens on screen that is needed to understand it. A video needs this.
Who they matter to
Deaf and hard of hearing people read what they cannot hear. Deafblind people depend on a descriptive transcript because they can neither hear the audio nor see the picture; they read it on a braille display, for example [3]. Transcripts also help people who take in text faster than speech, and anyone who wants to find or quote a particular passage.
What WCAG requires
| Content | Level A | Level AA |
|---|---|---|
| Audio only (podcast) | transcript (1.2.1) | – |
| Video only (no sound) | text or audio track (1.2.1) | – |
| Video with sound | descriptive transcript or audio description (1.2.3) | audio description (1.2.5), plus captions from Level A (1.2.2) |
For video with sound, then, a transcript on its own is only enough at Level A. At Level AA you need captions and audio description [2].
In code
The transcript belongs right by the player, as text on the page or as a clearly labelled link.
<figure>
<audio controls src="episode-12.mp3"></audio>
<figcaption>
Episode 12: Accessible forms ·
<a href="episode-12-transcript/">
Read the transcript
</a>
</figcaption>
</figure>
Common mistakes
- Uncorrected automatic transcription. Names, numbers and technical terms are often wrong.
- No speaker names, so it is unclear who said what.
- Missing sounds and visual content that are needed to understand it.
- Hard to find, on a separate page with no link at the player.
- Download only, for instance as a PDF that is itself inaccessible.
How Reviseberg handles it
Whether a transcript matches its recording is a question about content that no automated check answers, and Reviseberg does not listen to audio or watch video. What it checks on every crawled page: an audio recording with no transcript named beside it, and a silent video with controls and no text alternative, are reported (1.2.1); long text beside a player that does not say what it is becomes a question; and a video with neither audio description nor a text alternative is asked about (1.2.3). So 1.2.1 and 1.2.3 count as only partly testable by automation and stay "partial" until a person records a result with its method, such as "compared the transcript of every episode with the audio". Guided manual tests set out the procedure step by step, and a recorded result is valid for twelve months.
Related terms
Captions · Audio description · Braille display · Plain language · Manual accessibility testing