Accessible video captions are accurate, synchronized, complete text of a video's audio, including speaker identification and relevant sounds. They let deaf and hard-of-hearing viewers fully experience video content. This guide covers what accessible captions require, how WCAG applies, and the practical checklist for social video.
What Are Accessible Captions?
Accessible captions do more than transcribe dialogue. A truly accessible caption track includes:
- Every spoken word, accurately transcribed.
- Speaker identification when it is not obvious who is talking ([SARAH]:, [NARRATOR]:).
- Relevant non-speech audio: [laughter], [music], [applause], [door slams]. If a hearing viewer would get meaning from the sound, a deaf viewer needs it in text.
- Synchronization: captions appear and disappear in time with the audio, readable at a natural pace.
This is stricter than the burned-in style captions most creators use for TikTok and Reels, which often skip sound descriptions. Both have their place: styled captions for social feeds, full closed captions for accessibility-grade publishing.
Why Captions Support Inclusion
Around 430 million people worldwide live with disabling hearing loss (World Health Organization), and hundreds of millions more watch with sound off by choice or circumstance. Captions serve both groups with the same text.
Beyond the audience math, there is a legal and platform direction that only moves one way. Accessibility requirements keep expanding across regions and platforms, and brands increasingly require captioned deliverables from creators they work with. Learning accessible captioning now is future-proofing, not charity.
Caption Accuracy and Timing Standards
The WCAG (Web Content Accessibility Guidelines) standard that governs captions is Success Criterion 1.2.2 (Level A): captions must be provided for prerecorded audio content. The practical bar used by captioning professionals:
- Accuracy: 99% or better for compliance-grade work. AI drafts land in the mid-90s on clear speech, which is why human review is mandatory for this tier.
- Timing: captions synchronized to the audio, displayed long enough to read (generally a minimum of one second, maximum around seven seconds per caption).
- Reading speed: aim for 160 to 180 words per minute maximum. Faster than that and viewers cannot keep up.
- Line length: 32 to 42 characters per line, maximum two lines.
- No overlapping speech lost: when speakers overlap, represent both.
For social video, you will rarely hit every professional benchmark, and that is fine. The direction matters more than perfection: each step toward these standards makes your content more accessible.
How to Review AI Captions
AI captioning plus human review is the practical accessibility workflow for creators:
- Generate the draft in OneDrop's free AI caption generator.
- Correct every word. At this tier, "close enough" is not enough. Fix every misheard term.
- Add sound descriptions. Go through and add [music], [laughter], and speaker labels where needed.
- Check timing. Every caption should appear as its sound begins and clear before the next begins.
- Sound-off test. Watch the entire video on mute. If you can follow everything, including tone shifts signaled by sound, the captions work.
For content with legal exposure (education, healthcare, government, enterprise training), consider professional human captioning services. AI plus review covers the creator tier; regulated content deserves the professional tier.
Accessibility Checklist for Social Videos
- Captions present on every video with speech
- Spelling and names verified
- Speaker changes labeled where unclear
- Key sounds described ([music], [applause])
- Text readable: high contrast, adequate size, two lines max
- Captions positioned clear of platform interface overlays
- Sound-off watch-through completed
- SRT file uploaded for long-form (YouTube, courses) so viewers can toggle and screen readers can parse
FAQs
What does WCAG require for video captions?
WCAG Success Criterion 1.2.2 (Level A) requires captions for prerecorded audio content in synchronized media. In practice this means accurate, timed text covering speech and meaningful sounds.
Are AI-generated captions accessible?
They can be, once reviewed. Raw AI output typically falls short on accuracy (mid-90s vs the 99% professional bar) and omits sound descriptions and speaker labels. AI plus a proper human review pass meets the practical bar for most creator content.
Do social media captions count as accessible?
Styled burned-in captions help enormously and are far better than nothing, but they usually skip sound descriptions and speaker IDs. For full accessibility on long-form content, pair them with a proper SRT caption track.
How do I make my TikToks more accessible?
Add accurate captions (not just TikTok's plain auto-captions), keep text large and high-contrast, describe important sounds in your caption text where practical, and avoid flashing effects that can affect photosensitive viewers.
Make your videos more accessible with OneDrop.