Home/What Is Closed Captioning?
Verified July 2026

What Is Closed Captioning? Definition, Laws (ADA/WCAG/CVAA), Formats & Standards (2026)

Closed captioning is on-screen text that reproduces a video's spoken dialogue, speaker labels, and non-speech audio cues (music, laughter, [door slam]) — encoded so viewers can toggle it on or off, unlike open captions which are burned into the video. Required for prerecorded web video under WCAG 2.1 Success Criterion 1.2.2 (Level A) and for live web video under SC 1.2.4 (Level AA); mandated for U.S. broadcast television under FCC Report and Order 11-36 (2012) and for federal agencies under Section 508 (revised 2018). Distinct from subtitles, which translate dialogue but omit sound-effect and speaker-ID cues.

WCAG SC 1.2.2 (Level A) requires captions for prerecorded synchronized media; SC 1.2.4 (Level AA) requires captions for live synchronized media. All numbers below are sourced from primary law texts and standards documents — links in the Methodology & Sources section.

By VexaScribe Editorial · Published July 3, 2026 · Verified

Closed Captioning at a Glance

WCAG 2.1
SC 1.2.2 (Level A)
prerecorded video
42
chars/line max
Netflix TTSGuide, BBC
April 2026
DOJ Title II deadline
large state/local govs
4
major CC file formats
SRT · VTT · DFXP · SCC

Closed captioning is the accessibility-first sibling of subtitles. It covers not just dialogue but speaker identity and every non-speech audio event a Deaf or hard-of-hearing viewer would otherwise miss. In the United States, four legal frameworks drive caption obligations: the ADA (Titles II and III), Section 508 of the Rehabilitation Act (revised 2018), the CVAA and its FCC implementation (Report and Order 11-36, 2012), and WCAG 2.1 (June 2018) as adopted by the DOJ Title II Final Rule (April 24, 2024). WCAG 2.2 (October 2023) added no new caption criteria. State/local governments face compliance deadlines of April 24, 2026 (populations of 50,000+) or April 24, 2027 (smaller jurisdictions).

Methodology & Sources

Primary law & standards texts

Every legal claim on this page traces back to a primary source. We do not cite secondary explainers or law-firm blog posts as the basis for any date, deadline, or conformance level. Where case law is mentioned, we cite the docket and settlement date only.

Verification statement

All numbers verified July 2026 against primary sources listed above. Where a range is given for auto-caption accuracy (~85–92%), the range represents third-party benchmarks on clean single-speaker English; real-world accuracy varies with accent, domain vocabulary, and noise level. No dollar figures on this page are estimates — only cases with verifiable public settlements are cited.

Closed Captions, Defined

A closed caption track is a time-coded stream of text alternatives that renders on screen synchronized with a video's audio. Unlike a plain transcript, a compliant caption track includes three distinct categories of information: dialogue (what each person says), speaker identification (who is speaking, especially off-screen or when multiple speakers are audible), and non-speech audio cues (music, laughter, applause, sound effects such as [door slam] or [phone rings], and ambient sound relevant to comprehension).

The word “closed” refers to the user's ability to toggle the caption track on or off — the text is stored as an overlay stream separate from the video file. Open captions, by contrast, are permanently burned into the video pixels and cannot be turned off. Open captions guarantee visibility on any player but sacrifice user control and translation flexibility.

Closed captioning has a specific U.S. broadcast history. Line 21 encoding — the analog signal reserved for closed-caption data on NTSC televisions — was introduced by PBS in 1972. The Television Decoder Circuitry Act (Public Law 101-431, 1990) required all televisions 13 inches or larger sold in the United States after July 1993 to include a built-in caption decoder chip. That mandate is the reason today's streaming devices, smart TVs, and set-top boxes all natively support caption rendering.

In streaming distribution the term SDH (Subtitles for the Deaf and Hard-of-Hearing) is often used interchangeably with “CC.” Historically SDH was distinct: it referred to same-language subtitles that included non-speech cues but were delivered in the subtitle stream rather than the broadcast caption stream. On Netflix, Apple TV+, Prime Video, and Disney+, SDH tracks include speaker labels and sound effects; standard subtitle tracks omit them. For accessibility compliance purposes the two are treated equivalently.

Captions vs Subtitles vs Open Captions vs Audio Description

These four terms are routinely conflated. The distinctions matter for accessibility compliance because different laws demand different tracks.

TypeIncludes dialogue?Includes sound cues?Same language?User-toggleable?Primary purpose
Closed captions (CC)Accessibility for Deaf/HoH viewers
Open captions✗ (burned in)Same as CC but permanent
Subtitles✗ (no sound effects)✗ (different language)Translation for foreign-language viewers
Audio descriptionDescribes visual eventsAccessibility for blind/low-vision viewers

Source: WCAG 2.1 definitions §5; W3C Media Accessibility User Requirements.

Timing, Placement & Formatting Standards

These are the industry-standard technical specifications for producing compliant caption files. They come primarily from the Netflix Timed Text Style Guide and the BBC Subtitle Guidelines — the two most-cited public standards.

StandardValueSource
Max characters per line42Netflix Timed Text Style Guide (TTSGuide), BBC Subtitle Guidelines
Max lines per caption2Netflix TTSGuide, BBC
Reading speed15–20 chars/sec (160–180 wpm)WCAG-aligned, Netflix TTSGuide
Min caption duration1 secondBBC, Netflix TTSGuide
Max caption duration7 secondsNetflix TTSGuide
Min gap between captions2 framesNetflix TTSGuide
PlacementBottom-center (default)BBC, Netflix TTSGuide
Font/background contrast4.5:1 for normal textWCAG 2.1 SC 1.4.3

Sources: Netflix Timed Text Style Guide (TTSGuide), BBC Subtitle Guidelines, WCAG 2.1 SC 1.4.3.

Platform Requirements & Accepted Formats

What each major platform accepts on upload and whether its auto-captioning satisfies WCAG. Short answer: no auto-caption system on its own is WCAG-compliant, because compliance depends on accuracy, not presence.

PlatformAuto-captions?Accepted upload formatsPreferred formatNotes
YouTubeYes (auto, ~85–92% English)SRT, VTT, SBV, SCC, TTML, DFXPSRTAuto-captions NOT WCAG-compliant alone
VimeoNoSRT, VTT, DFXPSRTManual upload required
NetflixNoDFXP/IMSC 1.2IMSC 1.2 (strict)See Netflix TTSGuide
ZoomYes (live AI)VTT (post-meeting)VTTLive CC requires host enable
Microsoft TeamsYes (live)VTTVTTAuto-transcript separate
FacebookYes (limited)SRTSRTUpload during publish
InstagramYes (auto in-app captions)SRT (for uploaded videos)SRTReels have separate flow
TikTokYes (auto)SRTSRTAuto-captions ~85–92% accurate
LinkedInYesSRTSRTPublisher can upload manual
WistiaNoSRT, VTT, DFXPSRTCustom player supports both

Sources: YouTube caption formats, Netflix TTSGuide, vendor documentation for each platform.

YouTube auto-captions alone do not satisfy WCAG. Google's automatic speech recognition typically achieves 85–92% accuracy on clean, single-speaker English audio and drops to 60–75% on accented English, technical vocabulary, or noisy environments. WCAG 2.1 SC 1.2.2 and 1.2.4 do not include a numerical accuracy threshold, but courts and DOJ guidance increasingly treat quality — not presence — as the operative test. YouTube itself does not certify auto-captions as accessibility-compliant.

Compliance-grade workflow: use auto-captions as a first draft, then either edit them manually inside YouTube Studio or commission a human-reviewed pass from a professional service. The FCC 11-36 quality standards (accuracy, synchronicity, completeness, placement) provide a defensible benchmark for editorial review, even for content outside the FCC's direct jurisdiction. For live streams, the industry norm for compliance-grade captions is under 5 seconds of latency and greater than 95% accuracy — a bar that today's real-time AI captioners rarely clear on unrehearsed multi-speaker audio.

Caption File Formats Compared

Five caption file formats dominate 2026 workflows. Choice of format depends on delivery destination (streaming vs broadcast vs web) and styling requirements.

FormatExtensionMIME typeBest forStylingStandardization body
SubRip (SRT).srtapplication/x-subripMax compatibility, YouTube, editorsMinimal (bold/italic)De facto (SubRip 2001)
WebVTT.vtttext/vttHTML5 web video, styled captionsCSS + positioningW3C (May 2013 CG Report)
DFXP/TTML.ttml, .dfxpapplication/ttml+xmlNetflix, broadcast (IMSC 1.2 profile)Full XML stylingW3C TTML1/TTML2
SCC (Scenarist).scctext/plainUS broadcast TVEIA-608 codesEIA-608 (broadcast)
SBV.sbvtext/plainYouTube legacyNoneYouTube proprietary

Sources: W3C WebVTT specification, W3C TTML2 specification, Netflix TTSGuide (IMSC 1.2 profile), EIA-608 broadcast standard.

Companion authority pages for the two most widely used web formats: what is an SRT file and what is a VTT file.

Real Enforcement Outcomes

Captioning obligations only matter to the extent they are enforced. Five cases and one policy update anchor the current U.S. enforcement landscape.

NAD v. Netflix (2012, D. Mass. 869 F. Supp. 2d 196)

The landmark ADA Title III captioning case. The National Association of the Deaf sued Netflix over lack of captions on the Watch Instantly streaming service. The court held that a streaming-only service qualifies as a “place of public accommodation” under Title III. The consent decree required Netflix to caption 100% of its streaming library by end of 2014 and included attorney fee reimbursement. This ruling is the doctrinal foundation for applying the ADA to web video across all industries.

Harvard University & MIT (2015–2020)

NAD class actions alleging failure to caption online course videos, including edX and OpenCourseWare content. Both universities entered settlement agreements in 2020 committing to multi-year remediation plans requiring captions on all future public educational video. The Harvard and MIT settlements established that educational institutions cannot rely on “non-primary” content designations to avoid captioning obligations.

Domino's Pizza v. Robles (2019)

Not caption-specific but foundational for the private sector. The U.S. Supreme Court denied certiorari, leaving in place the Ninth Circuit's ruling that ADA Title III applies to websites and mobile apps of businesses that also operate physical locations. Robles cleared the path for the wave of subsequent web-accessibility lawsuits, including caption-focused claims.

DOJ Title II Final Rule (April 24, 2024)

The most consequential policy update in the current window. State and local governments face fixed compliance dates of April 24, 2026 (populations 50,000+) or April 24, 2027 (smaller jurisdictions). Enforcement pattern for missed deadlines is expected to follow compliance-audit and settlement-agreement conventions established under earlier Title II rulemaking rather than immediate individual civil suits.

Private-sector settlement pattern

Recent private-sector ADA Title III cases involving inadequate or missing captions — Peloton, Wynn Las Vegas, Winn-Dixie among them — have typically resolved through settlement agreements combining remediation timelines with plaintiff attorney fee reimbursement. Public settlement amounts vary widely and are frequently sealed; specific dollar figures are not reproduced here unless they appear on a court docket.

Auto-Captions vs Human-Reviewed Captions

The economic choice most publishers actually face in 2026.

Auto-caption accuracy. YouTube's automatic captions achieve approximately 85–92% accuracy on clean single-speaker English audio recorded in studio conditions, based on third-party benchmarks. Accuracy drops to 60–75% on accented English (Indian English, Scottish English, non-native speakers), on technical vocabulary (medical, legal, engineering domains), and on noisy audio (background music, ambient noise, low-quality microphones). Live auto-captioning on Zoom, Microsoft Teams, and Google Meet exhibits similar patterns. TikTok, Instagram, and LinkedIn in-app auto-captions land in the same 85–92% band on clean English.

Why WCAG compliance is not just presence. WCAG 2.1 SC 1.2.2 requires a “synchronized alternative for media” — a caption track that accurately reproduces dialogue and provides sufficient non-speech audio information. Both the DOJ and settlement-tracking case law increasingly interpret “alternative” to require adequate quality, not merely the presence of an SRT file. Garbage-in-garbage-out auto-caption tracks fail the spirit and letter of the rule.

Human transcription pricing. Rev (consumer tier) charges roughly $1.25 per minute for verbatim transcription and $1.50 per minute for captioning with timestamps. 3PlayMedia enterprise captioning is roughly $2.75–3.50 per minute with full styling support. VITAC and CaptionMax handle broadcast-tier work at custom enterprise pricing. Hybrid pipelines (AI transcription followed by human review) typically fall in the $0.50–1.00 per minute range with 24–48 hour turnaround.

Related VexaScribe guides: AI vs human transcription and how accurate is Whisper.

Live Captions — WCAG 2.1 SC 1.2.4 Level AA

WCAG 2.1 Success Criterion 1.2.4 requires captions for all live synchronized media — webinars, live-streamed conferences, real-time internal broadcasts, live-streamed lectures, streaming live sports and news re-delivered over IP. It is a Level AA criterion, so any organization citing “WCAG AA compliance” commits to captioning live content. Many small organizations quietly skip this obligation.

Latency expectation. Industry norm for compliance-grade live captions is under 5 seconds of end-to-end latency between speech and caption display. FCC 11-36 (2012) sets equivalent expectations for regulated broadcast-derived IP video.

Realtime CART. Communication Access Realtime Translation is the human-stenographer service tier. CART writers use steno machines and typically deliver 98%+ accuracy at 200–300 words per minute with under 3 seconds of latency. Pricing runs $100–200 per hour for scheduled events; on-demand rates are higher.

AI live captioning. Google Live Caption (Android and Chrome), Otter Live, Microsoft Live Captions, and Zoom AI Captions produce captions in real time using cloud ASR. Accuracy is roughly 85–92% on clean single-speaker English; drops with accents, technical vocabulary, and cross-talk. Suitable as a first-line accommodation and for internal meetings; not sufficient on its own for compliance-critical live public events.

Federal live web events. Section 508 (revised 2018) requires captions for federal agency live-streamed events. Every US federal agency operates a caption vendor contract for this purpose.

Limitations & Edge Cases

Media alternative for text. WCAG 2.1 recognizes an exception: if a video is itself an alternative rendering of a text document already present on the page, and the page identifies the video as such, captions are optional. This exception is narrower than typically assumed and rarely applies to standalone marketing or educational video.

Foreign-language content. Captions in the source language (a same-language transcription) and translated subtitles (localization into a different language) are distinct products. WCAG obligations attach to the source-language track. Multilingual publishers commonly ship a same-language SDH track plus separately localized subtitle tracks for each language they support.

User-generated content platforms. UGC platforms have historically relied on DMCA-style safe harbors, but proactive captioning obligations are growing. The EU Accessibility Act enforcement window began June 28, 2025, and DOJ 2024 guidance clarifies that public-facing UGC surfaces on government sites are not exempt from Title II obligations.

Live vs prerecorded conformance. SC 1.2.4 (live) is Level AA, one conformance rung higher than SC 1.2.2 (prerecorded, Level A). Some organizations conform to Level A only and formally skip live captioning; that choice needs to be documented in an accessibility statement, and it excludes them from claiming Level AA or Level AAA compliance.

Quality, not just presence. Courts and regulators increasingly demand caption accuracy. An SRT file full of “[music]” placeholders and mis-transcribed nouns is not defensible under WCAG or Title II. Editorial review is the difference between a caption file and a compliant caption file.

Generate WCAG-aligned captions from any audio or video with VexaScribe →

Frequently Asked Questions

What is the difference between closed captions and subtitles?

Closed captions include speaker IDs and non-speech audio cues (music, laughter, sound effects) and assume same-language dialogue for viewers who cannot hear the audio. Subtitles translate dialogue for foreign-language viewers but typically omit sound effects and speaker labels. In streaming (Netflix, etc.) the terms are often used interchangeably as "CC" and "SDH" (Subtitles for the Deaf and Hard-of-Hearing).

Are closed captions required by law in the US?

Depends on who you are. Broadcast TV: yes, under FCC rules dating to 1993 and reinforced by FCC Report and Order 11-36 (2012). Federal agencies: yes, under Section 508 (revised 2018). State/local governments: yes, under DOJ Title II Final Rule (published April 24, 2024), with compliance by April 24, 2026 (large governments, 50,000+ population) or April 24, 2027 (small governments). Private businesses: case law under ADA Title III makes WCAG 2.1 Level AA the de facto standard.

Do YouTube auto-captions meet ADA requirements?

No, not on their own. YouTube auto-captions are approximately 85–92% accurate on clean single-speaker English but drop to 60–75% on accented English, technical vocabulary, or noisy audio. WCAG expects synchronized captions of adequate quality — accuracy is an expectation, not just presence. Auto-captions are a starting point that must be reviewed and corrected by a human for compliance-grade output.

What is the difference between WCAG 2.1 and WCAG 2.2 for captions?

None for captioning specifically. WCAG 2.2 (October 2023) added no new caption success criteria. SC 1.2.2 (prerecorded, Level A) and SC 1.2.4 (live, Level AA) from WCAG 2.1 (June 2018) remain the operative rules. The DOJ Title II Final Rule (April 2024) adopts WCAG 2.1 Level AA as its baseline conformance target.

Are live captions required by law?

For US federal agencies, yes — under Section 508 (revised 2018) and general Section 504 obligations. For public web content of state/local governments after April 24, 2026 (large) or April 24, 2027 (small), yes — under the DOJ Title II Final Rule. For private businesses, WCAG 2.1 SC 1.2.4 (Level AA) is the de facto standard cited in ADA Title III lawsuits. For IP-delivered video that previously aired on US broadcast television, FCC Report and Order 11-36 (2012) requires live captions per the CVAA phase-in.

What file format should I use for closed captions?

SRT for maximum compatibility (YouTube, video editors, most platforms). WebVTT (.vtt) for HTML5 web video with native browser support and CSS styling. DFXP/IMSC 1.2 for Netflix and broadcast-quality delivery with strict styling requirements. SCC (Scenarist Closed Captions, using EIA-608 codes) for US broadcast TV workflows.

What is the difference between WCAG SC 1.2.2 and SC 1.2.4?

WCAG 2.1 Success Criterion 1.2.2 (Level A) requires captions for prerecorded synchronized media — any recorded video with audio. Success Criterion 1.2.4 (Level AA) requires captions for live synchronized media — webinars, live streams, and real-time video. SC 1.2.4 is a stricter, higher-conformance-level requirement that many small organizations skip.

Do I need to caption videos on Instagram, TikTok, or LinkedIn?

For accessibility compliance under ADA Title III (private businesses), best practice is yes for all published video. Instagram, TikTok, and LinkedIn each provide auto-caption features, but manual review is recommended for WCAG-compliance-grade output. Auto-captions typically achieve 85–92% accuracy on clean English audio and are not sufficient on their own to satisfy WCAG.