Does a video with only background music need captions? W3C's table

No. W3C's accessibility guidance says a video whose only audio is background music does not need captions, because there is no important audio content to convey. Captions are needed when the audio carries information a viewer must have to understand the video. For that case, captions on prerecorded video are required at WCAG Level A, and captions on live video at Level AA. Sume's caption job needs audible speech to work from, so the question to settle first is whether your clip has any.

This follows the "Does My Media Need Captions?" section of W3C's Captions/Subtitles page, read 2026-10-03.

What does W3C's decision table say?

The page walks through four kinds of media. Here it is as a table.

What counts as audio information you need to caption?

W3C frames it as a question: does the video have audio information that is needed to understand what it communicates? Spoken dialogue and narration clearly do. A sound effect that explains the scene, like an alarm, can. A decorative music bed does not. W3C also defines captions as the text version of speech and non-speech audio information, synchronized with the audio.

It separates captions from subtitles by language: captions for the same language as the spoken audio, subtitles for speech translated into another. Captions are an accessibility accommodation; subtitles in other languages are not directly one.

Are burned-in captions allowed?

W3C notes that most captions are closed, meaning viewers can hide or show them, while open captions are always displayed and cannot be turned off. A Sume caption job produces open captions: it burns the text into a captioned video. That is right when the video travels without its player, such as a repost or an ad, and wrong when a viewer needs to hide the text. If the standard you are held to is WCAG, check whether your delivery needs a toggle before you choose.

What should you do about automatic speech-to-text?

W3C is direct: automatically generated captions do not meet user needs or accessibility requirements unless they are confirmed to be fully accurate, and they usually need significant editing. Its example is a missing "not" that reverses the meaning of an oven instruction. When you caption with Sume and you know the spoken script, pass it as script_text: Sume keeps the speech-to-text timings and aligns the burned wording to your script, so the approved words appear. Then review a frame or two before publishing.

What does Sume need from the clip?

Speech-to-captions needs audible speech. A silent clip fails as caption_no_speech, with next_action: use_overlay_captions. That error is correct for a video-only clip where W3C says no captions are required, and the useful reply is to leave it uncaptioned. If the clip is silent but you still want on-screen words, say a price or a headline, pass authored cues with text, start and end; that is overlay copy, and it is not a substitute for captions of speech.

Each accepted standalone job for a clip up to 60 seconds is $0.20, per the video captions docs. Submit, then poll the job as Jobs and results describes.