Blendpipe homeSend a brief
Menu

Captions and audio description for video ads: rules and delivery

Audio description for videos and captions in ads: what FCC rules, WCAG 2.2, Section 508 and ad platforms require in the US, and which files to deliver.

Reader question
Do video ads legally need captions or audio description in the US, and how should a brand deliver them anyway?
A small puppet theater with a half-open red velvet curtain on a wooden table, where a drawn yellow cartoon lantern lights the empty stage.

No federal rule makes a typical 15- or 30-second spot carry captions or audio description. The FCC’s television captioning rule defines video programming to exclude advertisements of five minutes or less, and its audio description quotas count hours of programs on stations and cable systems, not commercials. Requirements reach ads by other routes: WCAG 2.2 for web and app video, Section 508 for federal agencies, the ADA rule for state and local governments, and some networks’ own ad guidelines. The practical answer is to deliver a caption file for every version, open captions where a placement cannot show closed ones, and a script that works by sound alone, so that a described version is short work.

This guide is for brand, legal and media teams and for agency producers who write the deliverable list. It sets out what each rule covers as of November 2026, using the eCFR, the W3C, Section508.gov, ADA.gov and the platforms’ own help pages, and then turns the answer into files: caption formats, open-caption versions, described mixes and the checks before they ship. It is general information, not legal advice; your counsel decides how a rule applies to a specific campaign.

What federal TV rules require for captions and audio description in ads

Federal television rules require captions on long-form advertising only, and they put audio description duties on distributors, measured in hours of described programs. A standard 15-, 30- or 60-second spot falls outside both.

The caption rule excludes ads of five minutes or less

The FCC’s captioning rule, 47 CFR 79.1, requires TV stations, cable and satellite distributors and the networks that supply them to caption 100% of new, nonexempt English and Spanish video programming. The definition does the work for advertisers. Paragraph (a)(10) says video programming “includes advertisements of more than five minutes in duration but does not include advertisements of five minutes’ duration or less.”

Paragraph (d)(6) adds a second exemption for “interstitial material, promotional announcements, and public service announcements that are 10 minutes or less in duration.” One statute cuts back into that exemption. Under 47 U.S.C. 611, any television public service announcement “produced or funded in whole or in part” by a federal agency “shall include closed captioning of the verbal content,” and a station is not required to supply captions that the announcement lacks.

Two practical consequences follow:

  • Long-form ads are programming. A 28-minute paid program is an advertisement of more than five minutes, so it is video programming under 79.1 and the stations and networks that carry it answer for its captions like any show. Expect them to ask for caption data with the master.
  • Federally funded PSAs carry their own captions. If a federal agency pays for or produces the spot, the captions travel with it; the station will not add them.

The same section sets caption quality standards that are a useful yardstick even where the rule does not apply. Paragraph (j)(2) asks captions to be “accurate, synchronous, complete, and appropriately placed”: matching the spoken words in order, conveying speaker identity, music and sound effects, appearing when the speech begins, and never blocking faces or featured text.

Audio description quotas count programs, not spots

47 CFR 79.3 defines audio description as “the insertion of audio narrated descriptions of a television program’s key visual elements into natural pauses between the program’s dialogue.” Its duties fall on distributors:

  • Stations affiliated with ABC, CBS, Fox or NBC and licensed in the top 90 Nielsen markets (as of January 1, 2023) must provide 50 hours of audio description per calendar quarter in prime time or children’s programming, plus 37.5 additional hours between 6 a.m. and 11:59 p.m.
  • Cable and satellite systems with 50,000 or more subscribers owe the same hours on each of the top five national nonbroadcast networks they carry.
  • Compliance is counted per channel and per quarter, and a described program may be counted no more than twice.

The section never mentions advertisements. It sets quotas in hours of programs for stations and distributors, and it places no duty on the brand whose spot runs in the break.

Online video follows television, not the other way round

The internet captioning rule, 47 CFR 79.4, covers full-length programming and clips that were shown on US television with captions and are then delivered over the internet. Its duties fall on the owners, distributors and providers of that programming, and it excludes consumer-generated media. A spot made only for social feeds or a brand site never aired on television, so this rule does not reach it.

RuleWho it bindsDoes it reach a typical 30-second spot?
47 CFR 79.1, TV captionsTV stations, cable and satellite distributors, networksNo: ads of five minutes or less are not "video programming"
47 U.S.C. 611, PSA captionsFederally produced or funded TV PSAsYes, for those PSAs only
47 CFR 79.3, audio descriptionTop-four affiliates in the top 90 markets; distributors with 50,000+ subscribersNo: quotas count hours of described programs
47 CFR 79.4, online captionsOwners and distributors of TV programming delivered onlineNo, for a spot that never aired on TV with captions
WCAG 2.2Whoever adopts it by law, contract or policyYes, for video on sites and apps held to it
Section 508 (WCAG 2.0 A and AA)Federal agenciesYes, for video an agency publishes
ADA Title II rule (WCAG 2.1 AA)State and local governmentsYes, on their own sites and apps, from April 26, 2027 or 2028
Network and platform termsAdvertisers buying that inventoryDepends on the terms; FOX's guidelines cite WCAG
General information as of November 2026, not legal advice. / Sources: eCFR, 47 CFR 79.1, 79.3 and 79.4; 47 U.S.C. 611 (govinfo); W3C WCAG 2.2; 36 CFR 1194 Appendix A, E205.4; ADA.gov Title II rule fact sheet; FOX Advertising Guidelines (October 2024); all checked November 2026.

Where accessibility requirements do reach video ads

They reach ads through the places the ads run and the organizations that buy them. Web pages and apps are measured against WCAG, federal agencies work under Section 508, state and local governments under the ADA Title II rule, and some networks write accessibility into their advertising terms.

WCAG 2.2: the success criteria that apply to video

The Web Content Accessibility Guidelines 2.2 became a W3C Recommendation on 12 December 2024. WCAG is a voluntary standard until a law, a contract or an organization’s policy adopts it, usually at Level AA, which includes every Level A criterion. These are the criteria a video ad on a web page or in an app meets or fails:

Success criterion Level What it asks What it means for an ad
1.2.1 Audio-only and Video-only (Prerecorded) A An alternative for media with only sound or only picture A silent animated spot needs a text alternative or an audio track with the same information
1.2.2 Captions (Prerecorded) A “Captions are provided for all prerecorded audio content in synchronized media” Every version with sound needs captions
1.2.3 Audio Description or Media Alternative (Prerecorded) A Audio description or a full text alternative At Level A, a timed transcript-style alternative can satisfy it
1.2.5 Audio Description (Prerecorded) AA “Audio description is provided for all prerecorded video content” At AA, a text alternative alone no longer satisfies it
1.2.7 Extended Audio Description (Prerecorded) AAA Pausing the video when the gaps are too short Impossible inside a fixed broadcast slot; possible in a web player
1.4.2 Audio Control A Audio that plays automatically for more than 3 seconds needs a pause, stop or volume control Applies to autoplaying spots on a page
1.4.3 Contrast (Minimum) AA Text contrast of at least 4.5:1; large-scale text 3:1; logotypes exempt A practical yardstick for supers
2.2.2 Pause, Stop, Hide A Moving content that starts automatically, lasts more than five seconds and sits beside other content needs a pause, stop or hide control Applies to autoplaying video in a page layout
2.3.1 Three Flashes or Below Threshold A Nothing flashes more than three times in any one-second period, unless below the flash thresholds Strobe edits and flash frames

The contrast criterion is written for text and images of text on web pages. Whether it governs a super burned into a video frame is a judgment for the accessibility lead, but 4.5:1 is a sound target for any legal line a viewer must read.

Section 508 for federal agencies

Federal agencies follow the Revised 508 Standards, which say electronic content “shall conform to Level A and Level AA Success Criteria and Conformance Requirements in WCAG 2.0” (36 CFR 1194, Appendix A, E205.4). The Section508.gov page on synchronized media, reviewed in October 2026, spells out what that means for video:

  • Synchronized media with sound and picture needs both captions and audio description.
  • Automatically generated captions do not meet the minimum standard for prerecorded media.
  • Subtitles, which translate dialogue for hearing viewers, “are not an acceptable method” for meeting the standard.
  • Players must offer caption and description controls at the same menu level as volume, and an alternate described file “must be labeled as the audio described version.”
  • Agency policy may require open captions.

A recruiting campaign or a public health message for a federal agency should plan captions and description from the first script.

State and local governments

The Justice Department’s ADA Title II rule makes WCAG 2.1 Level AA the technical standard for the websites and mobile apps of state and local governments. According to ADA.gov, an interim final rule published on April 20, 2026 moved compliance to April 26, 2027 for entities serving 50,000 or more people, and to April 26, 2028 for smaller entities and special districts. Tourism offices, transit agencies and public universities buy video advertising, and the video they publish on their own sites and apps falls under that standard.

Businesses open to the public

For private businesses there is no federal web accessibility regulation with a technical standard. The Justice Department’s guidance of March 18, 2022 says Title III covers businesses open to the public, including what they offer on the web, and points to WCAG and the Section 508 Standards as “helpful guidance.” It names videos without captions as a barrier and describes accessible video as “including synchronized captions that are accurate and identify any speakers.” The guidance itself has no legally binding effect, so counsel decides what standard a brand’s own site is held to.

Networks and publishers

Some ad guidelines bring accessibility in directly. FOX’s advertising guidelines, as of October 2024, say advertising must meet FCC, FTC and other legal requirements “including, without limitation, compliance with applicable accessibility guidelines such as Web Content Accessibility Guidelines (WCAG)” (FOX Advertising Guidelines, PDF). Paramount’s advertising standards, last modified in May 2025, ask that required supers be “clearly legible (appropriate size and contrast)” and hold for 3 seconds for one line, 4.5 for two and 6 for three (Paramount advertising standards). The wider network review is the subject of the companion article on getting a TV commercial cleared.

What each ad platform supports for captions and description

Every major platform accepts captions in some form, but they differ in file format, in what the word “caption” means and in whether a described audio track can be attached. The table reflects each platform’s help or spec page as of November 2026.

Destination Captions Audio description Watch for
YouTube Caption files including SRT, SBV, WebVTT and TTML; SCC is “our preferred file format” (YouTube Help) A “Descriptive audio” track, for creators with Advanced features; an audio-only file “roughly the same length as your video” (YouTube Help) SRT and SBV must be plain UTF-8 with no style markup; the original or dubbed track in that language must exist first
Meta (Facebook Feed video) The Ads Guide lists video captions as “Optional, but recommended” (Meta Ads Guide) No described-track option found Meta’s help page on uploading caption files could not be read for this article; confirm the file naming in Ads Manager
LinkedIn video ads Captions “must be in SRT format,” text only, with no custom formatting (LinkedIn Help) None stated Font colors in the SRT are not supported
TikTok In-Feed ads On the spec page, “caption” means the ad’s text line, shown in white in a uniform font for non-Spark ads (TikTok Ads) None stated Burn in open captions if you need them; TikTok’s auto-caption help page did not render for this article
Broadcast TV (AICP spec) Optional embedded QuickTime caption track at 29.97 fps; first caption “pop-on” after one second or “paint-on” after four frames (AICP) A separate described version, as the network or distributor directs 23.98 fps caption tracks were not supported when the spec was issued
CTV publishers Each publisher’s sheet Each publisher’s sheet Ask before the edit; see the campaign delivery hub for publisher specs

Two points from the table change the edit. First, a platform that calls the ad text a “caption” gives you no subtitle track at all, so the only captions a viewer sees are the ones burned into the picture. Second, YouTube is the only platform in this set with a documented slot for a described audio track, and even there it depends on channel features.

Closed captions, open captions and subtitles: which to deliver where

Deliver closed captions as a sidecar file wherever the player supports them, burn in open captions where it does not, and treat subtitles as translation rather than captions. Most campaigns need all three in different cells of the matrix.

Closed captions are a separate track or file that the viewer turns on and off. Open captions are part of the picture and cannot be turned off; the WFA’s ad accessibility guide calls them “burnt in” and reminds advertisers to keep them “legible with high contrast to the background” (Ad Accessibility, access features). Subtitles translate dialogue for viewers who can hear it, and they leave out the sound effects and speaker labels captions carry. A transcript is the text without timing.

A decision rule for each placement

  1. The player reads a caption file. Deliver closed captions: SRT for LinkedIn, SRT, WebVTT or SCC for YouTube, an embedded caption track for broadcast per the AICP spec.
  2. The placement has no caption track, or its “caption” field is ad copy. Deliver an open-caption version, built as its own file.
  3. The placement often plays muted. Weigh open captions even where closed ones are possible, because a viewer has to find and enable a closed track. Do not deliver both on the same file, or the viewer sees the words twice.
  4. The audio is in Spanish. Caption in Spanish. English text over Spanish audio is a subtitle, not a caption.
  5. The spot will run on a federal agency’s site. Deliver closed captions with a player that exposes them; Section508.gov notes that agency policy may also require open captions.

Formats and what each one carries

Format What it holds Where it is used
SRT (SubRip) Cue numbers, timecodes and plain text YouTube, LinkedIn, most social and web players
WebVTT (.vtt) Timing, text and positioning; styling limited to bold, italic and underline on YouTube Web players, YouTube
SCC (Scenarist) Broadcast captions based on CEA-608 Broadcast delivery and YouTube, which prefers it
Embedded track Captions inside the QuickTime master Broadcast masters under the AICP spec

What good captions contain

Use the FCC’s quality paragraph as the house standard even where the rule does not bind you. Captions should match the spoken words in order, without paraphrasing unless timing forces it. They should name speakers when it is not obvious, mark music and sound effects that matter, begin when the speech begins and leave when it ends, and stay on screen long enough to read.

Placement matters more in ads than in programs, because ads are dense with text. Captions must not cover the legal super, the price, the web address or the packshot. If supers sit in the lower third, move the captions to the top for those seconds, and check the crop of every vertical version; the cutdowns and bumper ads guide covers how 9:16 and 4:5 crops change what survives in frame. TikTok’s spec page notes that its safe zone depends on the ad’s dimension and caption length and offers downloadable safe-zone files.

Audio description for videos: standard, extended and integrated

Audio description is narration that tells a listener what the picture shows. In an ad it either fits into the gaps around the voiceover or is written into the voiceover from the start, and the second route is cheaper and more reliable.

WCAG defines audio description as “narration added to the soundtrack to describe important visual details that cannot be understood from the main soundtrack alone.” The FCC’s definition adds where it goes: into “natural pauses between the program’s dialogue.” There are four ways to build it:

Type How it works Fits a fixed TV slot?
Standard Narration placed in the natural gaps in voiceover and dialogue Yes, if the gaps are long enough
Extended The video pauses so narration can run (WCAG 1.2.7, Level AAA) No; only in a player that can pause
Adaptive The mix is rebuilt from stems to make room, for example by lowering music under the narration; the WFA guide describes this route Yes, with a remix
Integrated The description is written into the original script, which the WFA guide calls “accessible by design” Yes; there is no separate version to traffic

What needs description in an ad

Describe what a listener would miss and needs: the product and brand when they appear only as picture, any on-screen text that the voiceover does not speak (offer, price, web address, legal line), actions that carry the message, and visual jokes. The WFA guide notes that UK rules require audio description to “read out all on-screen text”; US rules set no such list, but it is a sensible default. Do not describe what the soundtrack already makes clear.

How described versions are delivered

  • Broadcast and cable. A described version is usually a separate mix or a separate file. Ask the network or distributor how it wants described ads trafficked, and whether a separate version needs its own identifier.
  • YouTube. Upload an audio-only described track under “Descriptive audio” if the channel has Advanced features; otherwise publish a separate described video.
  • Web players. Offer a second audio track or an alternate file, labeled as the audio described version, as Section508.gov describes.

Fitting description into a 30-second spot: a worked example

Measure the time the voiceover leaves free, take a margin off each side of every gap, and multiply the usable seconds by the speaking rate. If the description does not fit, rewrite the voiceover; do not squeeze narration into the music.

The steps:

  1. Lock the voiceover and time it against picture.
  2. List every gap with its start and end.
  3. Subtract a margin at each end of each gap so the narration does not collide with the voice.
  4. Multiply each usable gap by the narration pace and round down to whole words.
  5. List what must be described, shot by shot, and write lines that fit each gap.
  6. If an essential item has no gap, move it into the voiceover.

Example (fictional). Orlenna is an invented brand and the Ridge Bottle is not a real product. The 30-second hero spot has a 42-word voiceover read at 150 words per minute: 42 / 150 x 60 = 16.8 seconds of voice. That leaves 30 - 16.8 = 13.2 seconds without voice, made up of the plan’s 6 seconds of music-only moments and 7.2 seconds to spare.

Suppose the 13.2 seconds fall in four gaps: A from 0.0 to 4.0 s (4.0 s, ridge at dawn), B from 7.0 to 10.0 s (3.0 s, the drawn line becomes a chart), C from 17.0 to 19.6 s (2.6 s, back on the trail) and D from 26.4 to 30.0 s (3.6 s, packshot). Check: 4.0 + 3.0 + 2.6 + 3.6 = 13.2 seconds.

Take 0.5 seconds off each end of each gap: 4 gaps x 1.0 s = 4.0 s, leaving 13.2 - 4.0 = 9.2 usable seconds (A 3.0, B 2.0, C 1.6, D 2.6). At 150 words per minute the narrator speaks 2.5 words per second, so the gaps hold A 7.5, B 5.0, C 4.0 and D 6.5 words, rounded down to 7 + 5 + 4 + 6 = 22 words.

Lines that fit: A “Dawn breaks over a high rocky ridge.” (7 words). B “Steam rises, drawing a chart.” (5). C “Back on the trail.” (4). D “Packshot: the Ridge Bottle and logo.” (6). The kinetic type “Twelve hours” in shot 7 needs no description, because the voiceover already says “Twelve hours hot.”

If the end card also carried a web address that the voiceover does not say, gap D would need more than 6 words. The fix is in the script: add the address to the voiceover and shorten another line.

Where audio description fits in a 30-second spotTimeline of the fictional Orlenna 30-second spot. Eight shots run across the top. Voiceover occupies 4.0 to 7.0 seconds, 10.0 to 17.0 seconds and 19.6 to 26.4 seconds, 16.8 seconds in total. Four gaps remain: 0 to 4, 7 to 10, 17 to 19.6 and 26.4 to 30 seconds. After a half-second margin at each end, the usable description windows hold 7, 5, 4 and 6 words at 2.5 words per second, 22 words in all. Picture Voice Description 12345678 3.0 s7.0 s6.8 s 7 words5 words4 words6 words 051015202530 s Voiceover Description window Margin, 0.5 s
Example (fictional): the Orlenna 30-second spot with the voiceover timed at 150 words per minute and a hypothetical gap layout. Description windows are the gaps minus 0.5 s at each end, at 2.5 words per second. / Basis: Blendpipe's fictional example campaign and the script timing calculator's default example.

The arithmetic gets harder as the spot gets shorter. The script timing calculator gives a rough budget of 30 words for 15 seconds and 12 words for 6 seconds at 150 words per minute. A 15-second cutdown that uses most of its 30 words leaves almost no gap, and a 6-second bumper leaves none, so short versions depend on the integrated route described next.

Writing a spot that works with sound off and with sound only

The cheapest accessible version is the one designed in. A voiceover that names the product and says what matters, and supers that are large, high-contrast and on screen long enough to read, make captions and description simple add-ons rather than new edits.

These script and design rules come from the sources above and from how disclosure law already treats ads:

  • Speak every super that carries meaning. A price, an offer, a web address or a claim qualifier that appears only as text is lost to a listener. The FTC’s Endorsement Guides define a clear and conspicuous disclosure in terms that fit accessibility well: if a claim is made in both picture and sound, the disclosure belongs in both, and one presented “simultaneously in both the visual and audible portions” is “more likely to be clear and conspicuous” (16 CFR 255.0(f)).
  • Judge legibility by the actual audience. The same section judges a visual disclosure by “its size, contrast, location, the length of time it appears,” and an audible one by “volume, speed, and cadence.” One of its examples is a television ad aimed at older consumers, where the disclosure is evaluated from their perspective, “including those with diminished auditory, visual, or cognitive processing abilities.”
  • Name the brand and product in the voiceover. A logo reveal is silent to a listener.
  • Do not let color carry meaning alone. The WFA guide advises against using “colour alone to communicate meaning,” for example a green tick against a red cross with no words.
  • Give supers contrast and time. Aim for 4.5:1 against the background, the WCAG 1.4.3 ratio, and hold each line long enough to read; Paramount’s 3, 4.5 and 6 second rule for one, two and three lines is a usable floor.
  • Avoid flash sequences. Keep strobe effects and flash frames under WCAG 2.3.1’s three flashes in any one second.
  • Place music-only moments where the picture carries the message. They become the description gaps, so put them under the shots a listener would otherwise miss.

Writing this way also helps the platform versions. A spot whose voiceover carries the message survives muted autoplay with captions, and a spot whose supers are spoken survives a listener with the screen out of sight.

Adding accessible versions to the deliverable matrix

Treat captions and description as rows in the deliverable matrix, agreed before the edit. That means one caption file per version, open-caption video files only for the placements that need them, and described mixes only for the lengths and channels where they will run.

Deliverable Format One per Notes
Closed caption file SRT or WebVTT; SCC or an embedded track for broadcast Version and language Timed to the final mix, named to match its video file
Open-caption version Separate video file Placement that cannot show closed captions Its own file name; for TV, ask whether it needs its own identifier
Described mix Audio file or full mix Length and language that will run described Mixed to the same loudness target as the main mix
Described video Separate video file Placement that cannot take a second audio track Labeled as the audio described version
Transcript Text Spot published on a web page The Level A alternative under WCAG 1.2.3

A described mix is still a commercial mix, so it must meet the same loudness target as the main version; the guide to loudness for TV and streaming ads sets out which targets are law and which are publisher practice. The narration is also a new voice performance. If the narrator is union talent, the session and usage terms apply as for any voiceover, which rights and usage explains.

Example (fictional). The Orlenna matrix is 3 lengths (30, 15 and 6 seconds) x 4 ratios (16:9, 9:16, 1:1 and 4:5) x 2 languages (English and Spanish) x 1 call to action = 24 deliverables, with 22.0 hours of edit and quality check at the versioning matrix calculator’s defaults.

Accessible rows added to it, as an illustration:

  • Caption files: one per deliverable, in the language of its audio = 24 files. At the calculator’s 10 minutes of quality check per file: 24 x 10 = 240 minutes = 4.0 hours.
  • Open-caption versions: suppose the media plan runs the 9:16 and 4:5 versions in feeds that usually play muted: 3 lengths x 2 ratios x 2 languages = 12 extra video files. At 45 minutes of edit plus 10 of quality check: 12 x 55 = 660 minutes = 11.0 hours.
  • Described mixes: the 30- and 15-second spots in both languages = 2 x 2 = 4 mixes. At 55 minutes each: 4 x 55 = 220 minutes, about 3.7 hours. The 6-second bumpers get none, because they have no gaps.

Added time: 4.0 + 11.0 + 3.7 = 18.7 hours, for a total of 22.0 + 18.7 = 40.7 hours. These are the calculator’s default rates, not Blendpipe prices or estimates.

The versioning matrix calculator runs this count for your own matrix. The deliverables page lists caption files among the standard handover items.

Automatic captions and AI audio description: where they help and where they fail

Speech recognition and AI drafting tools produce a first pass quickly, but every source that discusses them asks for human review. Section508.gov goes further and says automatic captions alone do not meet the federal minimum for prerecorded media.

YouTube’s own help page on automatic captions says they “might misrepresent the spoken content due to mispronunciations, accents, dialects, or background noise,” and tells creators to “always review automatic captions and edit any parts that haven’t been properly transcribed” (YouTube Help). The WFA guide says the same of any automatic captions: “you’ll need to verify the accuracy of them.”

Ads expose the weak points of automatic tools:

  • Invented words. Brand names, product names and coined terms are exactly what speech recognition gets wrong.
  • Claims and numbers. “Twelve hours” captioned as “12 ours” is a typo on a claim the legal team approved.
  • Music and sound. Automatic captions rarely mark the music cue or the sound effect that sells the joke.
  • Judgment in description. A tool can describe what is in a frame but cannot decide which visual detail the spot depends on.

Where they help is the first draft: a transcript to correct, timings to adjust, a list of shots to describe. AI audio description for videos works best on the same terms, as a drafting step that a writer checks against the picture and the brief. If the narration uses a synthetic voice, choose a stock voice or get written consent before cloning any real person’s voice. Blendpipe lists AI-assisted caption and description drafts in the provenance log with the person who reviewed them, as its AI practices page describes.

Quality checks before the accessible files ship

Check accessible files against the final mix and the actual player, not against the script. Most caption errors come from late voiceover changes that never reach the caption file.

  1. Words. Every caption matches the final voiceover and dialogue word for word, with product names and numbers spelled as approved.
  2. Timing. Captions start when speech starts and leave when it ends. For broadcast, follow the AICP rule: the first caption pops on after one second of content or paints on after four frames.
  3. Sound information. Music, meaningful sound effects and speaker names are marked.
  4. Placement. No caption covers a super, a face, the price or the packshot; captions move to the top where supers sit low.
  5. Reading time. No caption flashes by faster than it can be read.
  6. File format. SRT files are plain UTF-8 with no styling for YouTube and text only for LinkedIn; broadcast tracks run at 29.97 fps.
  7. Description coverage. Every unspoken super with meaning is voiced, and no narration overlaps the voiceover.
  8. Description mix. The described mix meets the same loudness target and peak limit as the main mix.
  9. Labels and names. Described files are labeled as audio described versions, and every caption file carries the name of the video file it belongs to.
  10. Supers. Contrast, size and time on screen are checked on a phone and a TV, and no sequence breaks the three-flash limit.
  11. Player test. Each file is played on the real platform with captions on and with the described track selected.

The delivery specs checklist includes the caption line for every file alongside the other delivery checks.

What to do next

Decide where the spot will run, then work backward from the strictest destination. The steps, in order:

  1. List placements and buyers. Note any federal agency, state or local government client, network with accessibility terms, and platform that cannot show closed captions.
  2. Write for sound off and sound only. Speak meaningful supers, name the product, and place music-only moments under shots a listener needs described.
  3. Time the gaps. Run the voiceover through the script timing calculator and map the gaps before the edit locks.
  4. Add the rows. Put caption files, open-caption versions and described mixes into the deliverable matrix and the schedule.
  5. Ask each destination. Confirm caption format and described-version handling with every network, publisher and platform contact before delivery.

Blendpipe is a production studio that combines animation, live-action footage, 3D graphics and AI-assisted tools, and captions are part of its finishing stage, as the post-production page describes. If you are planning a campaign, include your placement list and any accessibility requirements in the brief you send through the contact page, and they will appear as rows in the production plan.