What Is WCAG 1.2.4?
You are streaming a webinar. Broadcasting a town hall. Running a live Q&A. Whatever the format — if it has audio and it is synchronised media (video with audio), anyone who cannot hear the audio needs captions in real time.
That is what 1.2.4 is about. It extends the captioning requirement from prerecorded content (1.2.2) to live content. The same principle applies — spoken audio needs a text equivalent — but the technical challenge is different because there is no opportunity to prepare captions in advance.
This is a Level AA requirement, which means it sits in the zone that most accessibility laws — ADA, PSBAR, the European Accessibility Act — use as their benchmark.
| Field | Details |
| WCAG Criterion | 1.2.4 |
| Conformance Level | Level AA |
| Principle | Perceivable |
| WCAG Version | 2.1 |
| Official Reference | https://www.w3.org/WAI/WCAG21/Understanding/1-2-4 |
The Concept in Plain Terms
Live captions appear on screen in real time as someone speaks. They are not perfect transcripts — they may have a short delay, minor errors, or shorthand — but they give deaf and hard of hearing users access to the spoken content as it happens.
The key difference from prerecorded captions is that you cannot prepare them in advance. Live captions require either a human stenographer (CART — Communication Access Realtime Translation), an AI-powered live captioning service, or a platform that provides real-time captioning built in.
Accuracy matters. Low-quality auto-captions that mangle technical terms or proper nouns are better than nothing, but they need to be as accurate as practically achievable.
Who Does This Actually Affect?
- Deaf and hard of hearing users who rely on captions to access spoken content
- Users in noisy environments who cannot hear the audio clearly
- Non-native speakers who find reading easier than following rapid speech
- Users on muted devices — especially in professional or public settings
Live events often have the largest audiences. A webinar with 500 attendees where 50 cannot access the audio without captions is a significant accessibility failure — and in many sectors, a legal one.
Common Failures
No captions at all
The simplest and most common failure: a live stream with no captioning solution. The broadcast goes out, deaf users get nothing.
Captions switched off by default
A platform that supports live captions but has them off by default, with no obvious way for users to enable them, creates a barrier. Captions need to be available and accessible — not hidden in a settings menu.
Over-relying on low-quality auto-captions without review
Many platforms offer AI auto-captioning. For live content, accuracy can drop significantly — especially with accents, technical vocabulary, or poor audio quality. Using auto-captions without any quality management does not always meet the spirit of this criterion, even if it technically provides ‘some’ captions.
Post-event captions only
Adding captions to a recording after the live event is good practice but does not satisfy 1.2.4. The requirement is for captions during the live broadcast.
How to Fix It
Option 1: Use a platform with built-in live captions
Tools like Zoom, Microsoft Teams, Google Meet, and YouTube Live offer built-in live captioning. The quality varies. Test your chosen platform ahead of any major event — especially if speakers have accents, use technical terms, or if audio quality could be variable.
Option 2: Use a CART provider
CART (Communication Access Realtime Translation) involves a professional stenographer providing real-time captions, typically at 98%+ accuracy. This is the gold standard for live captioning. Cost is higher but appropriate for high-profile or high-risk events.
Option 3: Use a third-party live captioning service
Services like Verbit, 3Play Media, and Ai-Media offer live captioning combining AI with human correction. These bridge the gap between platform auto-captions and full CART.
Whichever approach you use:
- Test it before the event — do a run-through with the actual speakers
- Give speakers a vocabulary list if using AI captioning — especially proper nouns and technical terms
- Make captions visible and enabled by default
- Brief speakers to speak clearly, at a measured pace
Edge Cases Worth Knowing
- Audio-only live streams: If your live content is audio-only (a live radio broadcast, a phone-in), this criterion does not apply directly — it covers synchronised media. However, the spirit of accessibility still applies.
- Events with one speaker vs panels: Multi-speaker events are significantly harder to caption accurately, especially with overlapping speech. Allow extra lead time and quality planning.
- International audiences: If you are broadcasting to audiences in multiple languages, live captioning in the original language may not be sufficient for all users. This is a separate consideration from WCAG compliance.
How to Test for 1.2.4
Testing live captions requires attending or observing the live event:
- Confirm captions are enabled and visible before the broadcast starts
- Observe caption accuracy during the event — are they readable and substantially correct?
- Check whether captions are available by default or require user action to enable
- Note the caption delay — a delay of 3–5 seconds is typically acceptable; longer becomes disorienting
Why This Matters Beyond a Checkbox
Live events are often the most time-sensitive and socially important content you produce. A town hall about a policy change. A public health briefing. A company all-hands. These are exactly the moments where equal access matters most.
Adding live captions is also increasingly straightforward — most platforms have built-in tools. The barrier now is usually awareness and planning, not technology.
Quick Fix Checklist
- All live synchronised media (video with audio) has real-time captions
- Captions are enabled and visible by default, not hidden in settings
- Captioning solution has been tested before the live event
- Speakers have been briefed on speaking pace and clarity
- For high-stakes events, CART or a professional captioning service is used
Frequently Asked Questions
Does this apply to a pre-recorded webinar replay?
No — replays are prerecorded content and fall under 1.2.2. However, if you record a live event and make it available afterwards, you should add or clean up captions before publishing the recording.
How accurate do live captions need to be?
WCAG does not specify an exact accuracy percentage. The standard is ‘captions’ which implies a reasonably accurate real-time text equivalent. The practical expectation is that they should be accurate enough to be genuinely useful — which means errors should be occasional, not constant.
What if we cannot afford a CART provider?
Platform auto-captions are a reasonable starting point for smaller events and lower-stakes content. For significant public-facing events, budget for CART or a professional captioning service. The cost of inaccessibility — legal, reputational, and human — generally exceeds the captioning cost.
We use Teams/Zoom — are we covered?
Potentially, but do not assume. Test whether captions are on by default, whether they are accessible to all participants (not just the host), and whether the accuracy is adequate for your content. Platform defaults vary.