Accessible Training Video Production: Captions, Transcripts and the European Accessibility Act

The European Accessibility Act came into force in June 2025, and a good number of Irish organisations are still working out what it means for the video sitting inside their learning management system. The short version: if your training content is part of a service covered by the Act, or if you are a public sector body already bound by the Web Accessibility Directive, your video needs to be usable by people who cannot hear it, cannot see it, or cannot process it the way you assumed everyone would.

Most teams discover this after the fact. They have a library of forty or fifty modules built over five years, none of them captioned properly, and someone in legal has just asked a question nobody can answer. Retrofitting that library costs multiples of what it would have cost to build accessibility into the brief at the start. This piece is about doing it at production stage, where it is cheap, rather than as a remediation project, where it is not.

What the European Accessibility Act Actually Requires of Video

The Act itself does not contain a list of video specifications. It sets out functional accessibility requirements for products and services, and then defers to harmonised standards for the technical detail. In practice that means EN 301 549, the European standard for ICT accessibility, which in turn incorporates the Web Content Accessibility Guidelines (WCAG) 2.1 at Level AA.

So when people ask about european accessibility act video requirements, the honest answer is that they are asking about WCAG 2.1 AA applied to time-based media. That standard is specific, testable, and has been around long enough that no production company should be treating it as novel.

The Four Success Criteria That Matter

For pre-recorded training video with synchronised audio, four criteria carry most of the weight:

• 1.2.2 Captions (Prerecorded) , captions must be provided for all pre-recorded audio content in synchronised media. This is Level A, the baseline.
• 1.2.3 Audio Description or Media Alternative (Prerecorded) , either an audio description track or a full text alternative that conveys equivalent information. Also Level A.
• 1.2.5 Audio Description (Prerecorded) , at Level AA, the text alternative alone no longer suffices. You need actual audio description for visual information not conveyed in the main audio track.
• 1.4.3 Contrast (Minimum) , text and images of text need a contrast ratio of at least 4.5:1 against their background. This catches a surprising number of otherwise well-made corporate videos.

There is also 1.4.5 (Images of Text), which asks you to use real text rather than text baked into an image wherever the presentation can be achieved with text. That one interacts awkwardly with video, because every on-screen graphic in a video is technically an image of text. The workable interpretation is that your transcript and captions carry the textual content, and your on-screen graphics meet the contrast and sizing requirements.

Who Is Actually In Scope

Public sector bodies in Ireland have been covered by S.I. No. 358/2020 since 2020, so for them this is not new. What changed in 2025 is that private sector organisations providing certain services (e-commerce, banking, transport, e-books, telecoms, and audiovisual media services) came into scope, along with the products supporting them.

Where it gets interesting for L&D is the knock-on effect. If you sell software to a covered organisation, your customer-facing training content is increasingly being pulled into procurement questionnaires. And separately, employment equality law and reasonable accommodation duties apply to internal training regardless of whether the Act names you directly. A staff member with a hearing impairment who cannot complete mandatory compliance training is a problem that predates 2025.

Captions: Standards, Formats, and the Burn-In Decision

Captions are the single highest-value accessibility investment in accessible training video production, and also the one most commonly done badly. Auto-generated captions from a video platform are not compliant. They routinely mangle proper nouns, drug names, technical terms, and Irish place names, and they do not identify speakers or describe relevant non-speech audio.

What a Compliant Caption File Contains

A caption file that meets WCAG expectations includes the spoken dialogue accurately transcribed, speaker identification when more than one person speaks, and non-speech audio that carries meaning (an alarm sounding, a machine starting up, a warning tone in a safety demonstration). Timing should be tight to the audio, with a reading rate that does not exceed roughly 160 to 180 words per minute for adult audiences.

Line length matters too. Two lines maximum, around 32 to 42 characters per line, broken at natural linguistic boundaries rather than mid-phrase. Broadcast conventions here are a reasonable guide, and the BBC subtitle guidelines remain the most usable public reference for anyone specifying this in a brief.

Closed Captions vs Open Captions in Training

The closed captions vs open captions training question comes up in nearly every accessibility conversation, and the answer depends less on compliance than on distribution.

Closed captions live in a separate file (WebVTT or SRT) that the player renders on demand. The learner can turn them on or off, change the size, and in some players change the colour and background. That user control is what WCAG prefers, and it is what most LMS platforms expect.

Open captions are burned into the video frame. They cannot be turned off, cannot be resized, and cannot be restyled. The advantage is that they survive any player, any platform, and any download. If your training video is going to be dropped into a SharePoint page, emailed as an MP4, played on a factory floor screen with no player controls, or embedded in a system that does not support sidecar caption files, open captions guarantee the text is there.

The practical recommendation for most enterprise clients is both. Deliver a clean master with a WebVTT sidecar file for the LMS, and an open-captioned version for the edge cases. The incremental cost at production stage is small because the transcript work is already done. Producing the second version later means going back to a project file that may no longer exist.

A Note on Caption Placement

If you are burning captions in, the graphics need to be designed around them. Lower-thirds, name supers, and data callouts placed in the bottom third of frame will collide with open captions. This is a storyboard decision, not an edit-suite decision, and it is one of several reasons accessibility belongs in the brief.

Transcripts and Audio Description Workflows

Captions handle the audio. Transcripts and audio description handle everything else, and they are where most organisations are weakest.

The Transcript Is Doing More Work Than You Think

A full text transcript serves learners who use screen readers, learners who prefer to read, learners on a poor connection, and anyone who wants to search the content later. It also happens to be the single best thing you can do for the searchability of training content inside a corporate intranet, because most enterprise search tools cannot index audio.

A compliant transcript is more than the caption file pasted into a document. It should include descriptions of meaningful visual content, so that someone reading it without watching gets the same information. If your safety module shows a technician performing a lockout procedure while the voiceover talks in general terms, the transcript needs to describe what is actually being demonstrated.

When You Need Audio Description, and When You Don’t

Audio description corporate video Ireland is a service that gets over-specified and under-specified in roughly equal measure. Audio description is a separate narration track that describes visual information during natural pauses in the main audio.

You need it when significant information is presented visually without being spoken. Think of a process demonstration where the narrator says “as you can see here” and then goes quiet, or an animated diagram that builds on screen with no verbal explanation of what is being built.

You largely avoid needing it when the script is written to be self-describing. If your voiceover explains what is happening rather than pointing at it, the main audio track already carries the information, and you satisfy the criterion without a separate description track. This is a scripting decision worth making early, and it is by some distance the cheaper route.

Where audio description is genuinely required, there are two approaches. Standard audio description fits into existing pauses. Extended audio description pauses the video to allow longer descriptions, which requires player support and is rarely practical inside a standard LMS. Building natural pauses into the edit for description is the more reliable route, and again, that is a production-stage decision.

Graphics, Colour, and On-Screen Text

The visual layer is where accessibility and design brief most often pull against each other, and where a brand guideline written in 2019 quietly makes your training non-compliant.

Contrast Ratios in Practice

The 4.5:1 minimum applies to normal-sized text. Large text (18pt regular or 14pt bold and above) can drop to 3:1. Many corporate palettes fail this comfortably. A mid-grey brand colour on white, a light blue on a lighter blue background, or white text over a video plate with a bright sky in it will all fall short.

The fix is usually straightforward: a solid or semi-opaque backing plate behind text over video, a darker variant of the brand colour reserved for text use, or a repositioned graphic that sits over a controlled part of the frame. Agreeing these variants once, at the start of a series, gives you a template that stays compliant across every subsequent module.

Colour as the Only Signal

WCAG 1.4.1 says colour cannot be the sole means of conveying information. In training video that usually shows up in charts (red bar bad, green bar good), in process diagrams, and in tick-and-cross compliance scenarios. Add a label, a pattern, an icon, or a spoken reference so the meaning survives for a colour-blind viewer.

Flashing, Motion, and Text Dwell Time

Nothing should flash more than three times per second (WCAG 2.3.1). That rules out certain stylised transitions and strobing effects that occasionally creep into motion graphics packages.

On dwell time, WCAG does not set a specific figure for video text, but the practical standard is that on-screen text should remain long enough to be read twice at a comfortable pace. In a wcag compliant elearning video, that usually means three to five seconds for a short line, longer for a list. Editors under pressure to keep the runtime down will cut this too fine unless it is specified.

LMS Versus Intranet: The Same Video, Two Different Problems

Two organisations can commission an identical video, deliver it to identical standards, and end up with different compliance outcomes because of where it plays. The player is part of the accessible experience.

Inside an LMS

If you are packaging training as SCORM or xAPI, the video player is usually the one built into your authoring tool or LMS. Check three things before you commission anything.

1. Caption support. Does the player accept WebVTT sidecar files, and does it expose a visible captions toggle? Some older SCORM players only support captions baked into the authoring tool’s own timeline, which changes your deliverable.
2. Keyboard operability. Can a learner start, pause, seek, and adjust volume without a mouse? This is the player’s responsibility, not the video’s, but it is your compliance problem.
3. Multiple audio tracks. Most LMS players cannot switch between a standard audio track and an audio-described one. Where that is the case, you deliver two separate video assets and offer both, which affects your file naming, your storage, and your course structure.

Our guidance on planning training video programmes covers the SCORM packaging side in more detail, and the two decisions are closely linked. Accessibility deliverables should be agreed at the same time as the packaging spec, not afterwards.

On an Intranet or Video Portal

Intranet delivery, whether through SharePoint, a corporate video platform, or an embedded player, generally gives you better caption handling and worse control over context. The video may sit on a page whose surrounding content is not accessible, in which case your compliant video lives inside a non-compliant experience.

Practical points here: make sure the transcript is published as page text or a linked accessible document rather than a scanned PDF, give the video a descriptive title and text summary rather than a filename, and confirm the player does not autoplay with sound. Autoplay is a genuine barrier for screen reader users, whose assistive technology competes with the audio.

The Multi-Language Overlap

If you are producing training for multiple markets, your caption workflow and your localisation workflow should be one process. The English transcript is the source document for subtitle translation, for voiceover scripting, and for the accessible transcript in each language. Running them as separate projects is how organisations end up with a Polish subtitle file that does not match the Polish voiceover. Anyone working across healthcare or multilingual training content will recognise the problem.

Costing Accessibility at Brief Stage

The reason to do this at production is arithmetic. Accessibility built into an original production typically adds a modest percentage to project cost, largely in transcription, caption authoring, quality review, and a slightly longer graphics approval loop.

Remediating an existing library costs more per asset because you are working without project files, without the original script, and often without anyone who remembers what the video was supposed to say. Add the co-ordination overhead of chasing forty modules across three departments and you are looking at a project rather than a line item.

What to Put in the Brief

• Target standard, stated explicitly (WCAG 2.1 Level AA)
• Caption deliverables: WebVTT sidecar, open-captioned version, or both
• Transcript format and whether visual descriptions are required
• Audio description approach: self-describing script, standard AD track, or none, with the reasoning recorded
• Contrast-compliant colour variants from the brand palette
• Target player and its known limitations
• Languages, and whether captions and voiceover are both in scope
• Who signs off on accessibility, and against what checklist

That last point is the one teams forget. Accessibility review is a distinct approval step with a distinct reviewer, and if it is not scheduled it gets skipped.

Frequently Asked Questions

Do auto-generated captions satisfy the European Accessibility Act?

No. Automatic speech recognition typically achieves 80 to 95 per cent accuracy on clean corporate audio, which sounds respectable until you consider that the errors cluster around exactly the terms that matter: product names, clinical terminology, regulatory references, and Irish surnames and place names. WCAG requires captions to be accurate and synchronised. Machine output is a useful first draft that a human then corrects and formats, and that is how most professional caption workflows operate.

Does our existing training library need to be remediated retrospectively?

It depends on your sector and whether the content is still in active use. Public sector bodies in Ireland have obligations covering existing content under the Web Accessibility Directive. For private organisations newly in scope of the Accessibility Act, the pragmatic approach most take is to prioritise mandatory and compliance training first, remediate high-traffic modules next, and retire or rebuild the long tail rather than remediate it. Take legal advice on your specific position, then let that shape the production plan.

How much does audio description add to a training video project?

Less than most people expect, if the need is identified before scripting. Writing a self-describing voiceover script costs nothing extra and removes the requirement in many cases. Where a separate description track is genuinely needed, you are looking at description scripting, a recording session, and a second mix, plus the delivery of an additional video asset. The cost climbs steeply if it is added after the edit is locked, because pauses may need to be created and the whole timeline reflowed.

Should captions be burned in or delivered as a separate file?

Deliver a WebVTT sidecar file as your primary caption asset, because it gives learners control over display and works with the accessibility features in most modern players. Add a burned-in version where the video will be distributed outside a controlled player environment, such as social platforms, downloaded files, or screens without player controls. Requesting both at production stage is inexpensive; requesting the second one two years later is not.

Who should sign off on accessibility before a training video goes live?

Ideally someone independent of the production, with a written checklist covering caption accuracy, transcript completeness, contrast ratios, colour dependence, flash thresholds, and player behaviour. In larger organisations this often sits with a digital accessibility lead or with IT. Where no such role exists, the L&D owner can run the check with a documented checklist and a contrast-checking tool. What matters is that it happens as a defined stage with a named owner, and that the result is recorded.

Building It In Rather Than Bolting It On

Accessible training video is not a specialist genre. It is ordinary training video produced by people who asked the right questions at the brief stage: what does the learner who cannot hear this need, what does the learner who cannot see this need, and does the player we are delivering into actually support any of it.

Ask those three questions early and the additional cost is marginal. Ask them after forty modules have shipped and you have a remediation programme on your hands, with a budget line nobody planned for and a deadline set by somebody else.

If you are scoping a training programme now and want to get the accessibility deliverables right from the first module, we are happy to talk it through. Get in touch with One Productions for a conversation about your project, or a free quote if you already have the brief.

Written by Freddie, One Productions.