WCAG 1.2.4 Captions (Live): How to Make Live Video Accessible

WCAG 1.2.4 Captions (Live) requires live audio that conveys information to have captions in real time. This page explains when the criterion applies, what good live captions look like, how to provide them, and practical steps you can use today.

What WCAG 1.2.4 requires

The success criterion says live audio-only and live audio in synchronized media must be provided with captions so people who are deaf or hard of hearing can access the content as it happens. Captions must convey speech and relevant non-speech sounds and be presented in a way users can read while the event is live.

When this rule applies

• Live streamed webinars, presentations and conferences
• Video calls and meetings that are broadcast or recorded live
• Live news, sports, emergency announcements and press conferences
• Any live audio-only stream that communicates information

Key caption quality elements

Meeting WCAG 1.2.4 is not just about adding text. Captions must be useful. Check these elements.

• Accuracy — Words should match the speaker where reasonably possible. Correct names and technical terms.
• Completeness — Include all spoken content needed to understand the event.
• Speaker identification — Label speakers when multiple people talk. Example: JUDGE: or HOST (off mic):
• Non-speech information — Tag meaningful sounds: [applause], [sirens], [music].
• Readability — Short lines, punctuation, and no dense blocks.
• Latency — Captions must appear with minimal delay so viewers can follow along in real time.
• Placement and visibility — Captions must not be obscured by player controls or overlays.

Methods to provide live captions

Choose a method based on accuracy needs, budget, and event complexity. Each method has trade-offs.

Human stenographer (CART or stenotype)

Real-time stenographers deliver high accuracy and fast turnaround. Best for formal events, court, or situations that require near verbatim captions. Requires a skilled operator and an audio feed.

Remote captioning services

Captioners work remotely from a streamed audio feed. They are slower to set up than local CART but cost less for some events. Works well for webinars and broadcasts.

Respeaking

A trained respeaker repeats speech into speech recognition software, adding punctuation and speaker cues. It gives higher accuracy than raw ASR when the respeaker is skilled.

Automatic speech recognition (ASR)

ASR systems provide instant captions. Accuracy varies with audio quality, accents, and jargon. Use only when budgets or timelines rule out human captioners, or as a hybrid with human correction.

Hybrid models

Combine ASR with human editors who monitor the captions and correct errors in real time. This reduces cost while improving accuracy.

Practical step-by-step implementation checklist

1. Decide scope: who is your audience and what level of accuracy you need.
2. Choose method: CART, remote captioner, respeaking, ASR or hybrid.
3. Contract and SLAs: specify turnaround, accuracy targets, and fallback procedures.
4. Provide materials: scripts, speaker list, technical terms, pronunciations, slide deck.
5. Set up audio routing: clean feed to captioner or ASR engine. Use mix-minus for calls to avoid echo.
6. Integrate captions: use your player’s live caption API, WebVTT/SRT ingestion, or platform-native tools.
7. Run a dress rehearsal with captions active and fix timing, overlays and readability.
8. Offer a clear toggle or enable captions by default when possible, and publish instructions for users.
9. Record the event and capture captions for the post-event captioned recording and archive.

Captioner instructions template (copy and paste)

• Identify speakers on each new turn: HOST:, SPEAKER 2:.
• Transcribe verbatim when possible. For unclear audio use [inaudible] with a timestamp.
• Include non-speech sounds in square brackets: [laughter], [door closes].
• Keep line length under two short lines on typical players.
• Avoid overlapping captions; queue rapid speech as brief chunks.
• Use proper nouns and technical terms from the provided script list.

Technical notes (formats and integration)

Many HTML5 players accept WebVTT for live captions. Live captioning may use a streaming caption feed or player API to inject cues. If you use ASR from a vendor, confirm how their captions are delivered and whether timestamps and speaker tags are supported.

Testing and validation

Do both automated and human testing.

• Check sync: captions must match the spoken words within a small delay.
• Check coverage: all important speech and non-speech info are present.
• Run real-user testing with people who are deaf or hard of hearing.
• Observe readability on phones, desktops and TVs. Test with screen magnification.
• Test captions during common failure modes: background noise, crosstalk and fast speech.

Common mistakes and how to avoid them

• Relying solely on unedited ASR. Use a human reviewer for events where accuracy matters.
• Not routing the clean audio feed to the captioner. Use a direct feed, not a room microphone when possible.
• Hiding captions behind UI controls. Reserve screen space and test player overlays.
• Turning captions off by default for public accessibility events. Enable captions or document how to turn them on.

Decision framework: which method to choose

Quick rules of thumb.

• High-stakes events with technical language: use a human stenographer or hybrid with editors.
• Frequent low-budget streams: use ASR plus human spot-checks.
• Interactive meetings: remote captioner or platform-native live captions with a human backup.

Recordkeeping and policy

Include captioning requirements in event contracts and accessibility policies. Keep a log of caption provider performance, error rates you observed, and post-event feedback. Use this data to improve future events.

Wrap-up

WCAG 1.2.4 Captions (Live) is about giving live audiences readable, timely and accurate text alternatives. Plan early, pick the right captioning method, and test with real users. Small changes in audio routing and captioner instructions make a big difference.

Quick checklist

• Decide method and budget
• Provide script and pronunciation list
• Route clean audio to captioner or ASR
• Test for latency, accuracy and visibility
• Record captions for the post-event version