How to Add Captions to Videos: Complete Guide (2026)

Why Captions Matter — Beyond Accessibility

Captions are usually framed as a disability accommodation, but the audience that benefits is much larger:

• Deaf and hard-of-hearing users: roughly 466 million people globally (WHO, 2024). Without captions, video content is inaccessible.
• Cognitive disabilities and ESL viewers: users who process text faster than rapid speech, or whose first language differs from the audio, comprehend captioned video at significantly higher rates.
• Sound-off mobile users: ~85% of Facebook video views happen muted (industry data). Captions multiply average view-completion by 2–3× on social feeds.
• SEO: Google indexes caption transcript text. A video with English captions ranks for the words spoken in it; without captions, the video is opaque to search.
• AI search visibility: generative engines (Google AI Overviews, ChatGPT, Perplexity, Claude) use caption transcripts to summarize, cite, and recommend videos. No captions = no citation.
• Legal compliance: the ADA, European Accessibility Act (EAA, effective June 2025), and AODA all mandate captions for public-facing video. The 2017 NFB v. HBO case established caption obligations for streaming platforms.

How Captions Work: WebVTT and the <track> Element

On the open web, captions are delivered as WebVTT files (the W3C standard) and attached to HTML5 video via the <track> element. Each cue has a start and end timestamp, the text, and optional speaker IDs and styling cues.

<!-- Bad: <track> without kind="captions" defaults to "subtitles" -->
<video controls>
  <source src="demo.mp4" type="video/mp4">
  <track src="demo.en.vtt" srclang="en" label="English">
</video>

<!-- Good: explicit kind="captions" with English as the default track -->
<video controls>
  <source src="demo.mp4" type="video/mp4">
  <track
    kind="captions"
    src="demo.en.vtt"
    srclang="en"
    label="English"
    default
  >
  <track
    kind="captions"
    src="demo.es.vtt"
    srclang="es"
    label="Espanol"
  >
</video>

A WebVTT file is plain text. Cues include speaker IDs, sound effects in brackets, and music descriptions:

WEBVTT

NOTE Product launch demo - English captions

1
00:00:00.500 --> 00:00:03.000
[upbeat instrumental music]

2
00:00:03.000 --> 00:00:06.500
SARAH: Welcome to the 2026 product launch.

3
00:00:06.500 --> 00:00:08.000
[applause]

4
00:00:08.500 --> 00:00:12.000
SARAH: Today we're shipping three things
you've been asking for.

5
00:00:12.000 --> 00:00:14.500
[door slams off-screen]

6
00:00:14.500 --> 00:00:17.000
MIKE: [speaking French] Where is the demo?

7
00:00:17.000 --> 00:00:19.000
SARAH: Right on cue.

How to Check Whether Your Videos Have Proper Captions

Captions fail silently. A page can pass every visual QA check while shipping videos with no caption track, with a track marked kind="subtitles" instead of captions, or with auto-generated captions that mangle proper nouns. Use these checks:

• Greadme deep scan — flags every <video> element without a <track kind="captions">, plus tracks with the wrong kind attribute or missing srclang/label.
• Greadme crawler scan — runs the same checks across every page on your site, so you can audit a full library of videos in one pass.
• Greadme AI visibility analyzer — shows whether AI engines are extracting your video transcripts (a strong downstream signal that captions exist and are well-formed).
• Sound-off playback test — mute the video and try to follow it using only the captions. If you cannot follow speakers, sound effects, or scene cues, the captions are incomplete.
• Screen reader test — VoiceOver, NVDA, and JAWS expose caption tracks via the player's controls. Confirm the captions menu lists your tracks and that the default track is correct.
• WebVTT validation — the W3C's WebVTT validator catches malformed cues, broken timestamps, and overlapping cues.

9 Rules for High-Quality Captions

Common Caption Mistakes and How to Fix Them

Captions vs Subtitles vs Transcripts vs Audio Description

These four media accessibility artifacts are often confused, but each one solves a different problem and is required by a different WCAG criterion.

FAQ

What is the difference between captions and subtitles?

Captions are written for viewers who cannot hear the audio. They include dialogue, speaker IDs, sound effects, and music cues. Subtitles are written for viewers who can hear but do not speak the language — they translate the dialogue and assume the viewer can hear non-speech audio. WCAG 1.2.2 requires captions; subtitles alone do not meet the criterion.

Are auto-generated captions WCAG compliant?

On their own, no. Auto-caption services typically deliver 70–85% accuracy on conversational speech, drop sharply on proper nouns, brand names, and technical terms, and rarely include speaker IDs or sound effects. WCAG 1.2.2 requires captions to convey the audio content equivalently — a 70–85%-accurate track does not. Use auto-captions as a draft, then edit them.

What is WebVTT and why does it matter?

WebVTT (Web Video Text Tracks) is the W3C standard for time-synchronized text on the web. It is plain text, supports cues with timestamps, speaker IDs, sound effects, positioning, and styling, and is loaded via the HTML5 <track> element. Every modern browser supports it natively. Files use the .vtt extension.

Should social media videos use open or closed captions?

Open captions (burned into the video) are usually safer on social feeds because many platforms either auto-play with sound off or do not surface caption controls reliably. For long-form content on your own site, closed captions via <track> give viewers control (font size, language, on/off) and remain accessible to assistive technology. Many teams ship both — open for social, closed everywhere else.

Do captions help SEO?

Yes. Google indexes caption transcripts as text content associated with the video. A captioned video can rank for the words spoken in it; an uncaptioned video is opaque to search. Captions also feed video schema (VideoObject with a transcript property) and improve dwell-time signals.

Do AI search engines like ChatGPT and Perplexity cite videos?

They cite the transcripts. Generative engines do not watch video, but they readily ingest WebVTT files, video transcripts, and on-page descriptions. A video with a complete, well-formed caption track is significantly more likely to be summarized or cited in an AI answer than the same video with no captions. Without captions, the video effectively does not exist to AI search.

What is the legal exposure for shipping uncaptioned video in 2026?

Substantial. The ADA has been applied to digital video in dozens of US lawsuits (notably NFB v. HBO in 2017). The European Accessibility Act became enforceable in June 2025 and covers private-sector video on commercial sites. AODA (Ontario) and the Accessible Canada Act apply in Canada. In regulated industries (government, education, finance, healthcare), uncaptioned video is a documented compliance failure that triggers remediation orders.