Captions and Accessibility in Australia: WCAG, the DDA, and What You Actually Have to Do

Why WCAG is the standard people point to

WCAG is published by the World Wide Web Consortium and is the reference point almost everyone in Australia ends up using, because it is specific where broader accessibility principles are not. It sets out testable criteria at three conformance levels — A, AA and AAA — rather than leaving “accessible” open to interpretation.

The Australian Human Rights Commission has published guidance that points organisations to WCAG as the practical reference for accessible online content, which is a large part of why it has become the common benchmark here.

For a lot of organisations the question is settled by policy or contract rather than by first principles. Commonwealth government bodies have long worked to WCAG 2.0 AA, with WCAG 2.1 AA now the common baseline across government and increasingly written into procurement. If you sell to government, the standard often arrives through the contract.

The criteria that apply to audio and video

You do not need to read the whole of WCAG. For media, a handful of success criteria carry almost all of the weight.

For most organisations the practical reading is simple. Prerecorded video needs captions. Audio-only content needs a transcript. Both are Level A — the minimum conformance level, not an advanced target.

Live captioning at 1.2.4 is where cost rises sharply, and it is a genuine consideration for organisations that stream events. That is a different product from what we do; we are asynchronous only.

Captions and transcripts are not the same thing

These get used interchangeably and WCAG treats them as distinct.

Captions are time-synchronised text displayed with the video. They serve someone watching who cannot hear the audio, which means they need to convey not just dialogue but relevant non-speech sound — a doorbell, laughter, a phone ringing — where it matters to understanding.

Transcripts are a static text document of everything said. They serve someone who wants to read rather than watch, someone using a screen reader, and anyone who wants to search or quote the content. A transcript alone does not satisfy the captions criterion for video, and captions alone do not give you the searchable document.

The practical answer for most video is to produce both, because you get the transcript for free once you have transcribed the audio, and the caption file is a formatting step away.

Are auto-generated captions good enough?

This is the question that matters most in practice, and it deserves a careful answer rather than a marketing one.

WCAG does not set a numerical accuracy threshold. But captions that misrepresent what was said do not provide an equivalent experience, and the accessibility community's long-standing objection to unreviewed automatic captions — memorably nicknamed 'craptions' — is well founded. Captions full of errors can be worse than none, because they look like provision while leaving the viewer with a garbled account.

The realistic position is that modern AI transcription produces captions that are close, and a human skim closes the gap cheaply. Errors cluster predictably in proper nouns, technical terms, and passages with crosstalk or poor audio — which means a reviewer knows where to look.

Where the standard usually shows up

Government. Commonwealth agencies and most state and territory bodies work to WCAG through published policy, and accessibility conformance is a routine line item in government procurement. If you sell to government, expect the standard to reach you through the contract.

Education. Universities and training providers generally hold themselves to WCAG for teaching material, and uncaptioned lecture recordings are a recurring source of student complaints. Institutions typically have their own accessibility policy that sets the expectation internally.

Health and essential services. Where a video conveys something a person needs in order to use a service or make a decision, treating captions as optional is a hard position to defend to your own users, whatever the formal position.

Everyone else. Nothing about WCAG is scoped to large organisations, and the criteria that matter for video are the cheapest ones to meet. Smaller teams tend to skip captions because nobody has asked, rather than because the standard treats them differently.

The part that is not about compliance

About one in six Australians has some degree of hearing loss, so the audience captions serve is large. But the majority of caption users are not deaf. They are watching without sound on public transport, in an open-plan office, or in bed. Social platforms autoplay muted, and captioned video demonstrably outperforms uncaptioned video on engagement as a result.

Captions also give search engines something to index. Video is otherwise opaque to search; a transcript on the page is the only way the content of a video becomes findable.

At $0.02 per audio minute, captioning a library of a hundred one-hour videos costs $120. The compliance argument is real, but the business case usually gets there on its own.

Where the audio goes

One point specific to Australian organisations: video containing identifiable people is personal information, and often sensitive information. Recorded lectures, client testimonials, internal training featuring staff, and consultations all qualify.

Sending that video overseas for captioning is a cross-border disclosure engaging APP 8. It is a slightly absurd outcome to create a privacy problem in the course of solving an accessibility one, but it happens routinely because captioning is treated as a production task rather than a data-handling one.