ADA Title II meeting videoYouTube auto-captionsAudio descriptionClerk checklist
What 1.2.2 is actually asking for
W3C’s Understanding document for 1.2.2 describes captions that convey the audio content of prerecorded synchronized media. In clerk language, that is more than “words appeared on screen.”
A meeting caption track has to carry:
- Dialogue — the words as spoken, including the text of a motion, not a phonetic guess.
- Who is speaking — speaker identification when it is not obvious from the picture. A dais of eight people, a resident at the podium, and a staff member off-camera all need to be distinguishable.
- Meaningful non-speech sound — a gavel, an alarm, applause that changes what happened next. Not every cough. The sounds that are part of the record.
- Synchronization — captions timed to the picture so a reader can follow the vote as it happens.
That is a public-record job. Auto-captions are a first-pass transcript of whatever the microphone picked up.
Where YouTube and Zoom auto-captions break on civic video
Platform auto-captions are trained on general speech. A city council, school board, or planning commission is not general speech.
They routinely miss or mangle:
- Local names. Officials, staff, frequent commenters, and streets that do not appear in a national dictionary. “Councilmember Ruiz” becomes a different person. The resident who spoke against the variance is misnamed on the record.
- Ordinance and resolution numbers. “Ordinance 2024-18” lands as a date, a street number, or noise.
- The motion itself. The exact wording that was moved, amended, and adopted is the legal act. A paraphrase in the caption file is not the act.
- Who moved, who seconded, and the roll-call. Auto-captions often collapse a roll-call into a blur of surnames with no ayes and nays, or they miss the second entirely.
- Overlapping speech. A dais that talks over a resident, or two members conferring at a mic, produces a garbled line that no one can use six months later.
Zoom’s live auto-captions have the same failure modes, plus the extra problem that the live guess is often what gets saved with the cloud recording. That saved track is still an auto-caption. It does not become a 1.2.2 caption because the meeting ended.
None of this is a knock on using YouTube or Zoom. It is a statement about what those auto tracks are: a convenience overlay, not the accessibility program.
Do not rely solely on auto-captioning
Section508.gov’s synchronized-media guidance is direct: do not rely solely on auto-captioning for prerecorded media. Federal 508 practice is not Title II, but the technical point is the same family of requirements Title II points at through WCAG 2.1 AA.
Auto-captions also do not deliver WCAG 1.2.5. There is no second track describing the budget on screen, the map, or the slide someone is pointing at. Captions and audio description are different jobs. When a meeting needs audio description covers the talking-head stretches where the dais already said it, and the screen-shares where it did not.
Title II requires WCAG 2.1 AA for web content a public entity provides or makes available after its compliance date, including meeting video hosted on a third-party platform. Keeping the auto-caption box checked is not, by itself, conformance with 1.2.2.
Hosting is not captioning
You can stay on YouTube. You can keep Zoom as the recorder. Title II does not name a brand of player.
What you cannot do is treat the platform’s auto track as the caption program for the posted file.
Two success criteria get collapsed in a lot of clerk conversations:
| When | Criterion | What it covers |
|---|---|---|
| Livestream | 1.2.4 Captions (live) | Real-time captions on the stream |
| File you post | 1.2.2 Captions (prerecorded) | Accurate, synchronized captions on that file |
Live 1.2.4 captions are a starting point. They are not a cleaned 1.2.2 track. Names guessed in real time, ordinance numbers the engine invented, and “inaudible” lines still have to be right on the recording residents will search next year.
If you burn the live auto-captions into the YouTube upload and walk away, you have a posted file with an auto track. That is the 1.2.2 problem, not a 1.2.4 leftover.
A clerk’s video checklist separates the livestream row from the posted-file row so the two jobs do not get signed off as one.
What Aware Lens returns instead
Aware Lens, from AwareNow, Inc., takes the recording you already make — a YouTube link, a Zoom cloud recording, or an MP4 — and returns reviewed captions for the file you post.
That track is built as a public record: speaker labels, local names, the text of motions, punctuation a reader can follow. Names & Places lets you correct a surname or a street once and propagate it. You can push the reviewed captions back to the YouTube channel you already use. Residents get the full record — captions, transcript, audio description where visuals matter, a plain-language summary, and Ask Aware — on the Aware portal.
Export the caption file anytime. An Accessibility Conformance Report (VPAT) is available on request.
Auto-captions guess at the audio. 1.2.2 is the caption track on the posted file. Those are not the same thing.
Book a 20-minute demo. We will import one of your real recordings and hand back a reviewed caption track you can compare to the auto overlay.
The rest of the Title II video picture — dates, 1.2.5, archives — is on ADA Title II meeting video.
Sources
- W3C Understanding 1.2.2 (WCAG 2.1)
- W3C Understanding 1.2.4 (WCAG 2.1)
- W3C Understanding 1.2.5 (WCAG 2.1)
- Section508.gov synchronized media
- ADA.gov Title II web-rule fact sheet
Product explanation, not legal advice.F.R. Part 35 and current ADA.gov materials before they go live.