Autoplay avatar welcome video with sound: WCAG 1.4.2 rule

The rule in one paragraph

A welcome video that starts talking the moment the page loads is the textbook trigger for WCAG 1.4.2. The W3C Understanding page (read 2026-10-04) states it as a Level A criterion: if audio plays automatically for more than 3 seconds, the page must offer a mechanism to pause or stop it, or a mechanism to control its volume independently of the overall system volume. Muting the whole computer does not count.

An avatar welcome clip from Sume Avatar 1.0 is audio by design: it speaks your script. Every clip the talking-video endpoint accepts is at least 4 seconds of estimated duration, so any autoplaying-with-sound version of it is past the 3 second line.

Three ways to stay inside the rule

You do not have to give up the welcome clip. You have to choose how the viewer gets control.

• Do not autoplay with sound. Show the first frame and a play button. This is the simplest option and avoids the criterion's trigger entirely.
• Autoplay muted, with a visible unmute and pause control. Muted media is not playing audio, so pair it with captions or a text alternative for the speech.
• Autoplay with sound and a prominent pause/stop control that is the first thing a keyboard user reaches.

Why the first frame matters

If you choose click-to-play, the poster is the whole pitch. Sume can produce a first-frame still before you pay for the final render through avatar video previews, so you can use that still as the poster. The docs describe preview_image_url for that still and say preview stills are never captioned, which keeps burned-in text out of your poster.

For a muted autoplay variant, a clip you caption inline will show the spoken words on screen. Inline captions burn into the final MP4 from the script, and the docs note they are rejected above 60 seconds of estimated duration, which is a limit you rarely meet with a welcome message.

Compare the options

Pick by how much you want the clip to do on its own.

What to ship

Render the clip with captions on if it will ever play muted, keep the script as page text, and make the control a real button element with a visible label. Keep the clip short: the longer a voice speaks unprompted, the more the control matters. If the welcome message is one of several onboarding steps, one job per step keeps each clip small, as in the onboarding checklist post.

The criterion is about the page, not the file. Sume gives you an MP4; whether it autoplays, loops or waits for a click is your player's decision, so the fix lives in your markup.

What this does not cover

Passing 1.4.2 does not make the video accessible on its own. The script and captions answer other criteria, and motion has its own rule, which the looping-clip post covers. Browsers also limit autoplay with sound on their own terms, so a clip you expected to autoplay may need a click anyway; design for the click case first and treat autoplay as an enhancement.