Alt text for an avatar video poster: WCAG 1.1.1 in practice

What to write

Describe what the still shows in a short phrase that identifies the content, and label the video itself so a screen reader user knows what it is. The Understanding page for 1.1.1 (read 2026-10-04) is Level A and says text alternatives for time-based media need at least descriptive identification of the non-text content, which is a lower bar than full equivalence. A poster alt such as the presenter facing camera in a bright kitchen, with the video label naming the topic, does the job for a clip whose speech is also published as text.

Where the poster comes from

Sume's avatar video previews endpoint, POST /v1/avatar-video-previews, returns first-frame stills before the final render, with preview_image_url and scene_previews[]. Regenerate and generate-video calls follow. Preview stills are never captioned, so your poster has no text baked in and the alt text carries the description.

You already know the avatar, scene and product from the request, which is most of what an alt text needs. Write it from your inputs, then glance at the still to confirm.

Three parts, three texts

An embedded avatar video has more than one thing to label.

• The poster image: alt text describing the still.
• The video control or container: an accessible name saying what the clip is, such as a welcome from the product team.
• The speech: the script published as page text; the script post explains when that is sufficient.

Examples

Keep alt text short and factual.

Keep the alt in sync

If you regenerate the preview, the still can change. Store the alt text with the preview id and update it when you pick a new one. Alt text is metadata on your page, not something Sume stores for you. Pair the poster with the other text alternatives described in the first-frame previews post so the page is complete before the final render.