Captions, Transcripts and Getting Video Accessibility Right

Captions serve three purposes simultaneously, which is unusual: they are an accessibility obligation, increasingly a legal requirement, and among the most useful text you can add to a video page.

Captions and transcripts are not the same thing

• Captions are time-synchronised and appear over the video. They should include speaker changes and meaningful non-speech sound.
• Subtitles assume the viewer can hear, and translate dialogue only.
• A transcript is the full text as a document, not synced. It is what search engines and readers use.

You want captions and a transcript. They do different jobs, and publishing one does not satisfy the need for the other.

What auto-captions miss

Automatic speech recognition has become genuinely good, and it still fails predictably in the places that matter most:

• Proper nouns — names, places, products: exactly the terms people search for
• Technical vocabulary — the words carrying the meaning
• Speaker attribution — usually absent, which makes group discussion hard to follow
• Non-speech audio — a laugh, a door, a shift in music, all invisible
• Punctuation — approximate, and it changes meaning more often than people expect

For accessibility purposes, uncorrected auto-captions frequently do not meet the standard. A pass over names and technical terms closes most of the gap in a few minutes per video.

A workable process

Generate automatically, correct selectively, publish both. Pulling the machine transcript takes seconds; the human pass is where the value is added, and it only needs to touch the words that matter.

The side benefit

Every accessible video is also a searchable one. The transcript serving a deaf viewer is the same text that makes the page findable — a rare case where the right thing and the profitable thing are identical.