Why a talking head is harder to caption than it looks
A principal welcome or superintendent update is often the simplest video you post. One speaker, a quiet room, clear audio. The hard part is what the speaker says. They name schools, staff, programs and acronyms that speech recognition has never heard. They say them quickly because they say them every day.
Auto-captions guess at those words. A wrong school name in a closure message is more than a typo. It can send a family to the wrong building or keep a student home who should be in class. Words that tend to go wrong:
- School names. For example, a school named for a local family can come out as a common word or a different name.
- Staff names. New principals, coaches and board members are rarely spelled right.
- Program names and acronyms. For example, your district's name for its after-school program, or acronyms like IEP, AP and CTE, which can turn into unrelated words.
- Numbers. Dates, times, bus route numbers and phone numbers get split, merged or misheard.
The DCMP Captioning Key says proper names and specialized terms should be checked against authoritative references and spelled the same way throughout. For a district, the authoritative reference is your own list of school, staff and program names. If you are relying on the automatic track today, read are YouTube auto-captions ADA compliant? before your next message goes up.
What is on screen but never said
The second problem is the picture. Captions cover the soundtrack. They do nothing for information that only appears on screen:
- A lower-third with the speaker's name and title
- A slate with the closure date or the make-up day
- A phone number, URL or QR code
- A slide with the schedule or a list of affected schools
- B-roll of a building, a flooded lot or an event that the speaker never mentions
The W3C's Understanding page for audio description says description fills existing pauses in dialogue with information about actions, scene changes and on-screen text. A closure slate nobody reads aloud is exactly that gap.
What WCAG 2.1 AA asks of a recorded message
The DOJ's ADA Title II rule adopts WCAG 2.1 AA for state and local governments, including public school districts. For a recorded message on your channel or website, three points matter:
- Captions (WCAG 1.2.2). Recorded speech needs accurate, synchronized captions. See caption quality standards for what accurate means in practice.
- Audio description (WCAG 1.2.5). It is needed when the picture shows something the soundtrack doesn't. If the speaker says everything that appears on screen, the video usually doesn't need a separate description track. The full test is in when is audio description required?
- Live town halls (WCAG 1.2.4). Live audio needs captions, and YouTube's live auto-captions alone usually fall short. The replay is recorded video and needs accurate captions like any other upload. See livestream captions.
Only Level A and AA criteria are required. The Title II deadline is April 26, 2027 for public entities of 50,000 people or more, and April 26, 2028 for smaller ones and special districts. These are the dates as extended in 2026. This post explains what the rules say. It isn't legal advice, so check your situation with your district's counsel.
Last year's welcome video
Old messages are not archived just because they are old. The archived content exception applies only when all four conditions hold: the video was made before the deadline, it is kept only for reference, it hasn't changed since, and it sits in an area clearly labeled as an archive. A welcome video still linked from a school's homepage fails that test. The details are in the archived video exception guide.
Social media is a separate question. The DOJ's small entity compliance guide says making every past social media post accessible may be impossible for many state and local governments, which is why posts made before the deadline are exempt. The DOJ's rule overview gives the example of a 2017 social media post by a city's sanitation department about delayed trash pickup after a snowstorm, close to a snow-day closure post. But neither DOJ nor HHS said whether YouTube counts as social media.
| Message | Where it lives | How to treat it |
|---|---|---|
| Uploaded to YouTube before your deadline | Only on your channel | Probably exempt (unclear) |
| Any message, any age | Embedded on a current school or district page | In scope: caption it and check for on-screen-only information |
| Any message someone asks for | Anywhere | Make an accessible version, either way |
| Uploaded after your deadline | Channel or website | In scope |
Urgent messages: write the script so it describes itself
Closures, weather delays and safety updates come with pressure to post in minutes. They are also the messages where a wrong caption does the most harm. A parent who relies on captions needs the right date and the right school as much as anyone else, and at the same moment.
The fastest fix happens before you press record. If the soundtrack carries every date, number and link, the picture holds nothing the audio doesn't. That is the simplest way to keep a message from needing separate audio description. Build these habits into the script:
- 1
Say who is speaking
Open with name and title. That covers the lower-third.
- 2
Say the date and the day
Say “Tuesday, January 12,” not “tomorrow.” Viewers watch the video hours later.
- 3
Name each school in full
Read the list of affected schools aloud, even if it is also on a slide.
- 4
Read numbers slowly
Say phone numbers digit by digit, and say them twice.
- 5
Say the link in words
Never rely on a QR code alone. Say where to go, such as the district website and the page name.
- 6
Mention what the b-roll shows
If you cut to footage of an unplowed road or a closed gym, say what it shows and why it matters.
Hello, I'm [full name], [title] of [district name]. All [district name] schools will be [closed / on a two-hour delay] on [day of week], [month] [date]. This includes [school name], [school name] and [school name]. [Meals / child care / activities]: [what families need to know]. For updates, go to [website name] and select [page name]. You can also call [phone number, digit by digit]. Again, that number is [phone number]. We will share our next update by [time] on [day].
Want to test a message you already posted? Answer a few questions about it here.
Fixing captions in the YouTube caption editor
YouTube Help says you can change both the text and the timestamps of your captions in YouTube Studio. It also says that editing auto-generated captions doesn't change the automatic track: a new caption track with your revisions is created when you use Duplicate and edit. Your script is the answer key.
- 1
Open the automatic track
In YouTube Studio, open the video's subtitles and choose Duplicate and edit on the automatic track.
- 2
Check names against your list
Go through every school, staff and program name. Spell each one the same way every time.
- 3
Check every number
Compare dates, times, phone numbers and web addresses to the script, character by character.
- 4
Add speaker names
When two people speak and the picture can't show who is talking, the DCMP Captioning Key puts the speaker's name in parentheses on its own line. For example, (Principal Rivera).
- 5
Fix timing
Use the two timestamp boxes next to each caption line so captions appear with the words, not seconds before or after.
- 6
Publish and play it back
Watch the first and last minute with captions on and confirm your corrected track is the one showing.
For a two-minute message, the whole check takes about as long as watching it twice. For the basics of uploading files, see how to add captions to YouTube videos.
A routine for a one- or two-person office
You don't need a big process. You need a few things set up once and a short check every time.
| When | What to do |
|---|---|
| Once, this week | Build a name list: every school, principal, program and acronym as you want it written. Keep it open next to the caption editor. |
| Once, this week | Save the script template and agree with your superintendent that anything shown on screen gets said aloud. |
| Before each shoot | Run a shot check: the lower-third matches what is said, no date or number appears only on screen, any QR code has a spoken link. |
| Before publishing | Duplicate and edit the automatic track. Check names, numbers, speaker labels and timing. |
| Urgent messages | If a message must go up before you can check it, fix the captions right after, starting with dates and school names. |
| Once, this month | Publish a way for anyone to request an accessible version of an older video. The request form template is a start. |
| Once, this month | Scan your channel and run the website video check to find older messages still embedded on current pages. |
Sources
- W3C WAI
- U.S. Department of Justice, ADA.gov
- U.S. Department of Justice, ADA.gov
- YouTube Help (Google)
- Described and Captioned Media Program (DCMP)
- Described and Captioned Media Program (DCMP)
This is a reading of the rules, not legal advice. Something out of date? Tell us and we'll fix it.