Accessibility
Accessible video and animation: captions, transcripts and motion
Accessible video requires more than automatically generated captions. What matters is whether users can perceive all relevant information, control playback themselves and operate the experience without a mouse. Plan captions, audio description where needed, a suitable transcript and an accessible player together. For animation, also address controllable movement, a sensible response to motion preferences and the avoidance of dangerous flashing.
The most efficient starting point is a media inventory: which content is recorded, which is live, and which communicates information exclusively through pictures or sound? This turns a broad demand for “accessible video” into specific production tasks. The examples below are hypothetical. They show how to organise meaningful checks from briefing to approval, without equating technical accessibility with complete legal compliance.
1. Classify requirements correctly
A WCAG 2.2 AA target includes all applicable Level A and AA success criteria. These include captions for prerecorded synchronised media at Level A and captions for live synchronised media at Level AA. At AA, audio description is required for relevant visual information in prerecorded synchronised media. The scope and exceptions of each criterion must be considered: a silent explainer video, for example, requires a different assessment from a conversation with pictures and sound.
The W3C’s WCAG 2.2 also distinguish requirements that are frequently confused. Disabling non-essential motion animation triggered by interaction belongs to criterion 2.3.3 at Level AAA. Respecting reduced-motion preferences is nevertheless good practice. EN 301 549 is not interchangeable with WCAG. For the Austrian legal context, see our guide to accessible websites in Austria: the BaFG has applied to covered services since 28 June 2025, not automatically to every website.
2. The brief determines later production effort
Before recording, identify which information will be communicated through each channel. In a hypothetical product video, the presenter says “Click here” while the pointer highlights an unlabelled icon. Without the picture, the essential instruction is missing. A better formulation is: “Open Settings and select Invoices.” This change also helps people who cannot identify the icon on a small screen.
The brief should define responsibilities, language versions, delivery formats and approval stages. Who checks terminology? Who describes important visual events? Who tests the embedded player? Allow pauses in the narration for descriptions and retain editable source files. If the team only discovers after editing that charts are explained exclusively through visuals, the audio, edit and running time may all need revision. Accessibility is therefore a production decision, not a single export format.
3. Captions also communicate the meaning of sound
Good captions do more than reproduce spoken words. They identify speakers when their identity is not clear from the picture and include meaningful sounds. In hypothetical safety training, “[warning alarm sounds]” may be essential because a machine must then be switched off. Irrelevant background sounds do not need to be described every time. The deciding factor is which acoustic information contributes to understanding what happens.
Automatic speech recognition provides a draft, not a reliable approval decision. Pay particular attention to negatives, numbers, names, dialect and technical terminology. “Do not release” must not become “release”. Also correct timing drift, awkward line breaks and excessively fast changes. Captions should not obscure important controls or on-screen instructions. A complete review with sound switched off is a useful acceptance test: is the message understandable, are speaker changes clear, and is safety-critical information preserved?
4. Deliver caption tracks and test their presentation
Separate caption tracks, for example in WebVTT, can often be switched on or off and offered in multiple languages. Captions burned into the picture remain visible and may be difficult to read in some output formats. The right approach also depends on the publishing channel. A translated dialogue track is not yet a complete accessible caption track if speaker identification and important sounds are missing.
Test the version that is actually published, not just the file in the editing application. One hypothetical failure: the German track exists but is labelled English in the player and is difficult to find. Correct the language assignment and menu labels, then test selection with a keyboard and screen reader. Check readability over light and dark scenes, on a narrow screen and in full-screen mode. Record the media version tested so that subsequent editing changes do not silently break synchronisation.
5. Transcripts and audio description serve different purposes
A transcript makes content readable independently of playback speed. It can include speakers, dialogue and relevant sounds; a descriptive transcript adds important visual information. This helps people looking up details, people with combined sensory impairments and users who do not want to play a video at that moment. Preferably provide it as structured HTML directly beside the media. A poorly accessible PDF merely moves the problem elsewhere.
Audio description, by contrast, communicates important visual information through sound as the content unfolds. In a hypothetical training film, it might need to explain that a status changes from “Draft” to “Sent” when this is not otherwise audible. A transcript beside the player is not a blanket substitute for audio description required at AA. Nor does it replace required captions. For prerecorded audio without video, however, a text alternative may be precisely the appropriate requirement.
If the original soundtrack already communicates all relevant visual information, an additional description may not be necessary. Assess the actual content rather than the video format. If natural pauses are insufficient, revise the production or provide a suitable described version. Extended audio description with additional pauses is classified as a separate Level AAA success criterion; it is not a blanket AA obligation.
6. Test the player as a complete interface
An accessible film remains inaccessible if its playback only works with a mouse. Test play, pause, volume, mute, captions, available descriptions, the timeline and full-screen mode. Controls need understandable accessible names and recognisable states. A screen reader should allow users to distinguish between a control that enables captions and one that disables captions already in use. Visual icons alone are insufficient.
Complete the entire journey using a keyboard. Focus must remain identifiable and, under WCAG 2.2 AA, must not be entirely hidden by author-created content. A timeline needs meaningful keyboard operation; where dragging is used, also consider the applicable AA requirement for a non-dragging alternative. Check that pop-up menus remain reachable and that focus returns appropriately when they close. Neither a standard player nor an external embedding service is automatic evidence of accessibility.
7. Live broadcasts need their own workflow
Live captions must be available during the event when the relevant AA criterion applies. A transcript published afterwards does not fulfil that purpose. Arrange captioning in advance, supply terminology and names, and test the complete transmission chain. This includes the event platform, the website embed and the devices the audience will actually use. Also decide who can respond if the captioning service fails.
During a hypothetical live demonstration, the presenter explains visible steps instead of referring only to the screen. This improves accessibility without implying a general AA requirement for live audio description. If the recording is subsequently made available on demand, assess it again as prerecorded media. Correct the live captions, carefully remove technical interruptions and add any required descriptions. Approval of the live format does not automatically approve the on-demand version.
8. Treat autoplay and automatic movement separately
Avoid automatically starting media wherever possible. If audio plays automatically for more than three seconds, WCAG 1.4.2 at Level A requires a way to pause or stop it, or volume control independent of the overall system volume. Muted autoplay avoids that particular audio issue but may still distract, cause nausea or fall under other criteria. A visible play button is often the clearer and more robust design.
For automatically starting movement, blinking or scrolling that lasts more than five seconds and appears alongside other content, a mechanism to pause, stop or hide it is generally required unless the movement is essential. The W3C explanation of Pause, Stop, Hide describes this distinction. Automatically updating information has a separate rule without that five-second threshold. A hypothetical background video behind a contact form therefore needs a different assessment from a brief loading animation.
9. Design reduced motion as an intentional alternative
Respect the operating system’s reduced-motion preference through prefers-reduced-motion and create a genuinely calmer version. A hypothetical page transition could move directly to the new content instead of using a large zoom effect. A parallax effect could be replaced by a static picture. Do not remove information, however: if movement explains a status change, that change must remain understandable through text, shape or another perceivable state.
The implementation should cover CSS animations and JavaScript-driven effects. Also check embedded components and content loaded later. A common mistake is to reduce animation duration drastically while leaving a disturbing jump visible. The correction is an alternative presentation, not merely a faster animation. Document reduced motion as a separate test configuration. It is valuable practice, but there is no universal standalone AA requirement under that name.
10. Do not approve flashing through rules of thumb
Wherever possible, avoid deliberate flashing at the design and source-material stages. WCAG 2.3.1 at Level A addresses flashes and thresholds; a simple counting rule cannot safely assess every flashing sequence. Image area, changes in luminance and saturated red can all be relevant. A pause button alone does not eliminate the risk if a problematic sequence plays before someone can react.
Also review rapid cuts, animated logos, screen recordings and supplied advertising inserts. For suspicious sequences, expert analysis or replacing the material is more useful than an unsupported assurance. A hypothetical trailer with a stroboscopic transition could use a calm cut instead. Record which version was assessed. A general warning about photosensitivity replaces neither the technical evaluation nor the necessary correction.
11. Base acceptance on reproducible user journeys
Define checks around concrete tasks. A person without sound should understand the essential statements, including warning signals. A person who cannot see the picture should be able to follow the process being described. A keyboard user should be able to enable captions, pause playback and leave the player. With reduced motion enabled, non-essential effects should disappear without losing information or orientation.
A useful test report records the timestamp, expected behaviour, observed failure and correction. For example: “At 02:14, a required step appears only on screen; the soundtrack does not mention it.” The retest then confirms the additional spoken instruction in that scene. Include the browser, device, assistive technology and media version. Automated checks can support the process, but they cannot reliably judge the subject-matter accuracy of captions or the completeness of a description.
12. Connect publication with ongoing maintenance
Treat the video, captions, transcript and descriptions as one publishing package. Changes to prices, procedures or functionality must be reflected in every affected version. Give customers and users an accessible way to report media problems. Internally, record who responds, who may replace files and how a correction is retested. This turns a problem report into an actionable task rather than a general complaint.
The W3C guidance on sustaining accessibility emphasises ongoing maintenance and organisational responsibility. Set your review frequency according to publishing activity and risk; WCAG does not prescribe a universal monthly or annual audit. Train editorial and production staff using recurring issues from your own workflow. A technical media assessment provides important evidence but does not replace a complete legal assessment of your service.
Frequently asked questions
Are automatic captions enough?
They are useful as a starting point, but not reliable as an unchecked final version. People need to verify meaning, terminology, speaker changes, relevant sounds and synchronisation.
Can a transcript replace audio description?
Not as a general rule. Where the AA requirement for prerecorded synchronised media applies, relevant visual information must be available audibly. An additional transcript addresses different user needs.
Is every animation prohibited at AA?
No. Relevant factors include automatic movement, duration, parallel content and flash thresholds. The criterion concerning disabling interaction-triggered animation belongs to AAA.
Does the BaFG cover every website containing video?
No. The deciding factor is the covered service, not the presence of a video. The WKO FAQ on the BaFG in e-commerce, in German, explain boundaries and exemptions.
Next step: assess media in its usage context
If you want to assess existing videos, a new player or an animation concept, start with a representative user journey. Zensations, founded in Vienna in 2010, can support you through an accessibility audit or its accessibility services. The objective is a prioritised basis for production, implementation and retesting, not a blanket guarantee of conformity or rankings.
Sources and editorial review
Content reviewed on 7 November 2026. Normative requirements and explanatory recommendations are distinguished throughout the article.
