Captions: text that follows the audio

This entry concerns time-based media. The central questions are what the audio communicates and when its text appears. It is not about the short label beneath a still photograph, and it does not treat every track called “subtitles” as containing the same information.

Read the text alongside its moment

Imagine a short fictional video for a model-building club. A presenter names a cardboard piece, a bell sounds, and the presenter says that the next demonstration is beginning. A caption review would consider the speech, the bell's relevance, and the moment each text segment appears.

A list of the presenter's words with no timing would answer a different question from a caption track shown with the video. In this example, the reader is examining how the text relates to the audio over time, not simply counting whether words exist somewhere.

An optional review ticket could be:

Clip: Model-club demonstration draft
Moment to inspect: Transition after the cardboard piece is named
Audio heard: Record the actual words and relevant sound.
Text shown: Record what appears at that moment.
Question: Does the text represent this part of the audio?
Status: Unresolved until the clip and text are compared.

The ticket is original. It is not a caption file, and the fictional clip does not establish exact timestamps that could be used in a real video.

Keep uncertain sounds uncertain

If a reviewer cannot hear a word clearly, they can flag that moment for the creator. A guessed name should not silently become a confirmed caption. A word list provided separately may help frame a question, but this glossary does not certify it as a faithful record of the recording.

The same caution applies to a sound description. In the fictional clip, the question is whether the bell matters to understanding that transition. The review ticket does not require every background noise to be written down or supply a universal rule for deciding.

Review with the creator's permission

A student making a class video might ask a family member to compare one section. Agree on whether the person is identifying issues, proposing wording, or actually editing the file. Do not replace a shared caption track while others are working on it without coordination.

If the video contains people or personal material, the captioning task does not itself authorize broader sharing. A short private question about one unclear moment may be enough.

A track label is not a quality certificate

This page does not claim that automatic captions are accurate, that a particular player supports a format, or that a video meets legal or technical accessibility requirements. Those require the actual media and relevant review. The W3C source supplies the terminology; the model-club ticket is an independent illustration of checking content and timing together.

Source

[1] Shawn Lawton Henry (editors). World Wide Web Consortium, Web Accessibility Initiative. Captions/Subtitles. Updated 2024-09-17; original publication date not shown. Accessed September 6, 2026. [1] uses only the terminology specified above; additional findings, recommendations, and source examples are not adopted.

Related guides

• Alternative text
• Text to speech
• Version history
• Clarifying question