Video and audio are now integral to modern websites, online shops and apps: product clips, explainer videos, webinars, podcasts. Yet a medium that works only through sound or only through images excludes people - deaf and hard-of-hearing users as well as blind and low-vision users. Germany is home to 7.9M (Federal Statistical Office, 2024) people with severe disabilities, and worldwide around 430M (WHO, 2024) people need care for hearing loss. On top of that comes the situational constraint: 85 percent (Digiday, 2016) of Facebook videos are watched without sound. Accessible media solves both - with captions, transcripts and audio description. This guide places the success criteria of WCAG guideline 1.2 in context, explains the difference between captions and transcripts and shows why automatic captions alone are rarely enough. Where your service stands today is clarified by a structured WCAG audit.
Why Accessible Media Has Become Mandatory
Since 28 June 2025, the German Accessibility Strengthening Act (BFSG) has required many private providers to make their digital services accessible. The BFSG transposes the European Accessibility Act (Directive EU 2019/882) and, for technical implementation, references the harmonized standard EN 301 549, which in turn references the success criteria of the Web Content Accessibility Guidelines (WCAG). Time-based media - that is, video and audio - are governed there by guideline 1.2. Anyone who embeds media files cannot avoid these criteria once the service falls within scope. Which BFSG requirements specifically apply depends on company size, sector and the type of service.
The situation is measurably patchy. The WebAIM Million Report (2024) found at least one automatically detectable WCAG violation on 95.9 percent (WebAIM Million, 2024) of the home pages analyzed, on average 56.8 (WebAIM Million, 2024) errors per page. Missing media alternatives are difficult to detect automatically, so the actual gap is likely larger than any automated count shows. At the same time, the benefit is broad: 80 percent (Verizon Media and Publicis Media, 2019) of people who use captions have no hearing impairment at all - they watch in loud environments, in the office or late at night with the sound off. Captions are therefore not a niche feature but improve the experience for everyone.
Accessibility and reach go hand in hand
WCAG Guideline 1.2 at a Glance
WCAG guideline 1.2 bundles all requirements for time-based media and is tiered by conformance level. At levels A and AA - the levels relevant for EN 301 549 and the BFSG - there are five success criteria. They build on each other: from audio-only or video-only without a second sensory layer, through synchronized captions, to audio description that makes information reserved for sighted users audible. The highest level, AAA, adds sign language among other things, but is usually not legally required.
| Criterion | Level | What is required |
|---|---|---|
| 1.2.1 Audio-/video alternative (prerecorded) | A | Audio-only needs a transcript; video-only (without sound) needs a text or audio alternative. |
| 1.2.2 Captions (prerecorded) | A | Synchronized captions for all speech and relevant sounds in prerecorded videos with audio. |
| 1.2.3 Audio description or full-text alternative | A | Essential visual information is conveyed - via audio description or an equivalent full text. |
| 1.2.4 Captions (live) | AA | Captions also for live content such as webinars or streams. |
| 1.2.5 Audio description (prerecorded) | AA | A full audio description for prerecorded videos, not just a text alternative. |
The distinction between terms matters. Captions are aimed at people who cannot hear the sound and therefore contain not only spoken language but also relevant sounds and speaker identification. These differ from translation subtitles, which only render the dialogue in another language and omit sound information. For accessibility under 1.2.2, the former, closed captions, are meant. How these criteria fit into the bigger picture of the most recent standard version is placed in context by our article on the new criteria in WCAG 2.2.
Creating Captions Properly - More Than Just Text
Good captions follow craft rules that go beyond mere transcription. They are in sync with the audio, stay on screen long enough to be readable and break at sensible points. A common guideline is around two lines and a reading speed in the range of about 160 to 180 words per minute, so that slower readers can keep up too. Crucial for accessibility is that not only dialogue but also action-relevant sounds - ringing, alarms, laughter - and, with multiple voices, speaker identification are captured. Captions thus follow the same principle as a good alternative text: whatever is conveyed through only one sensory channel gets an equivalent in text - as we explain in our article on accessible images and alt text.
Technically, closed captions are preferable to open ones: they exist as a separate track, can be turned on and off and remain searchable. On the web, the WebVTT format has become established for this, connected to a native video via the HTML track element. Open, burned-in captions, by contrast, are part of the image, cannot be switched off and are invisible to screen readers. During implementation it is also worth looking at the media player itself - it must be operable by keyboard, as we show in our article on keyboard navigation in web development.
<video controls>
<source src="tutorial.mp4" type="video/mp4" />
<!-- Closed captions as a separate track -->
<track kind="captions" src="tutorial-en.vtt"
srclang="en" label="English" default />
<!-- Audio description as a text track (fallback) -->
<track kind="descriptions" src="tutorial-ad-en.vtt"
srclang="en" label="Audio description" />
</video>
<!-- Link to the searchable full-text transcript -->
<p><a href="/transcripts/tutorial">Read the transcript of this video</a></p>Automatic captions are only a draft
Transcripts: The Underrated All-Rounder
A transcript is the complete text version of an audio or video item. For audio-only content such as podcasts it is the central alternative under criterion 1.2.1; for video, a well-considered full-text transcript can replace audio description at level A, provided it also describes the essential visual information. Unlike captions, a transcript is not tied to the timeline: it can be skimmed, searched, copied and read at leisure - an advantage for people with hearing or cognitive impairments as well as for users quickly looking for a specific passage.
A complete transcript - sometimes called a full-text alternative - contains not only the spoken language but also sound descriptions and all visually conveyed information needed for understanding. This makes a video usable even when neither sound nor image can be perceived, for example for deafblind people using a Braille display. As a welcome side effect, transcripts provide additional text context that search engines can index. Anyone offering transcripts as a PDF should also ensure their accessibility - our article on accessible PDFs and documents helps here.
Full text, not snippets
A transcript reproduces the entire content - spoken language, relevant sounds and descriptive cues about what happens on screen.
Searchable and copyable
Text can be skimmed and searched. Users find exactly the passage they need without playing the whole video.
Discoverable for search engines
Transcripts make the audio content available as indexable text and thereby improve the page's visibility on the side.
Usable for deafblind people too
Via a Braille display the content stays accessible when neither sound nor image can be perceived.
Timestamps for orientation
Time markers link text passages to the video and make it easier to jump to the desired point.
Basis for captions
From a clean transcript, synchronized captions can be derived efficiently - one work step, two results.
Audio Description: Images for the Ear
Audio description makes visual information audible that is essential for understanding and not already conveyed by the audio - such as a silent gesture, a scene change or on-screen text. In the original's natural speech pauses, an additional audio track describes what is happening. Criterion 1.2.3 (level A) allows either audio description or an equivalent full-text alternative; criterion 1.2.5 (level AA) then requires the full audio description. For low-barrier productions it pays to write the script from the outset so that important visual content is named aloud anyway - this lowers the later effort for a separate description.
In practice, time pressure is the biggest challenge: if the speech pauses are insufficient, WCAG provides for extended audio description at a higher level, where the video briefly pauses to make room for the description. For most BFSG-relevant services, standard audio description at level AA is enough. Anyone producing media anew should plan audio description, captions and transcript together rather than adding them afterwards. This integrated approach is part of our accessible web development and avoids costly rework.
An audio description succeeds when a person with their eyes closed can follow the action just as well as a sighted one - describing no more and no less than is needed for understanding.
Live Content, Third Parties and Player Embeds
Live captions under criterion 1.2.4 concern webinars, live streams and online events. They can be implemented through speech-to-text interpreting or through speech recognition with human correction in real time. Since live captions are more error-prone than those created afterwards, a recording is recommended for important content, which is then permanently available with corrected captions and a transcript. This turns the fleeting live format into sustainably accessible content.
Embedded third-party players are tricky. Many video hosting services do support caption tracks, but automatically generated captions are not accessible without review, and the third-party player's interface is not always fully usable by keyboard. Anyone embedding external content should also check its accessibility during the audit and name it in the accessibility statement if gaps remain. How such a statement is structured with the associated feedback mechanism is described in our article on the accessibility statement and feedback mechanism.
Do not forget the player and its controls
Anchoring Media Accessibility as a Process
Accessible media does not happen by chance but through a repeatable process. 90 percent (3Play Media, 2024) of organizations surveyed caption at least some of their content, but only 39 percent (3Play Media, 2024) consistently caption everything - it is precisely this gap that leads to non-accessible individual videos. A clear editorial rule helps: no video goes live without checked captions and a transcript, no podcast without a transcript. 66 percent (3Play Media, 2024) of organizations now define quality standards for caption accuracy - a good starting point for your own policy.
From our reference projects, it has proven useful to first inventory existing media and prioritize by reach: the most-watched explainer video first, rarely accessed legacy content later. Across 50+ (project experience) projects we have seen that a well-maintained transcript archive significantly lowers the effort for future captions, because caption tracks can be derived from it. The ongoing maintenance - producing new content accessibly, retrofitting legacy content - combines well with a regular re-audit and is part of our services.
- Inventory the media stock and prioritize by reach
- Embed captions as a closed WebVTT track rather than burning them into the image
- Always have automatic captions corrected manually
- Provide a searchable transcript for every audio and video item
- Plan audio description or a full-text alternative for essential visual information
- Test the media player and selection menus with keyboard and screen reader
- Name any remaining gaps in the accessibility statement
From a single video to a solid media strategy