Video captions are not a transcript. Plan both before you press record

Adding words to a video is not the same as making its information available to everyone. A viewer may need to follow the sound without hearing it, read without playing the video, or understand a chart they cannot see. Those are different editorial jobs. Plan for each before recording, rather than treating a caption button as the final step.

• Check captions against the finished audio, including relevant speakers and sounds.
• Offer a readable descriptive transcript, not just a dump of caption timestamps.
• Put essential visual information into the narration or an appropriate described version.

Three outputs, three different jobs

The W3C Web Accessibility Initiative's media guidance recommends providing captions and a separate transcript. They can share a checked text base, but their presentation and purpose differ:

• Captions: timed text alongside the video, carrying speech and the non-speech audio needed to understand it.
• Descriptive transcript: readable text covering that audio information plus essential visuals, organised so it works without playback.
• Audio description: essential visual information made available through sound, either within suitable main narration or through a described alternative.

Check the words against the final cut

WAI says automatic captions need accuracy verification; they can be a starting point, not a finished accessibility check. Listen through the actual export while reading the captions. A missing negative, a wrong number or a misheard name can alter the story. A script written before recording is not proof of what the finished soundtrack contains.

Its transcription guidance also covers relevant speaker identification and sounds. In an interview, identify a speaker when the viewer needs that information. In a device demonstration, label an audible warning if it explains what happens next. Not every background noise needs a label; include what carries meaning. Do not silently rewrite an interviewee's words into a cleaner argument.

Say what the picture is doing

Captions do not automatically explain a silent graphic. WAI's description guidance asks producers to convey visual information needed to understand the content, including important on-screen text. For a tutorial, replace a vague 'choose this one' with a spoken description of the control and action. For a chart, explain the relevant comparison, not merely that a chart exists.

This can be built naturally into some presentations and instructional videos. WAI calls it integrated description. It does not fit every format: a drama or densely edited sequence may need separate description. Plan pauses and shot duration early. Whether a separate track works also depends on the player, not just the file you export.

Make the text useful without the video

WAI's transcript guide explains how checked captions can become a transcript. Group short caption lines into paragraphs, identify speakers and add essential visual information missing from the soundtrack. Clearly distinguish editorial clarification from spoken words. Descriptive transcripts serve people who are Deaf-blind as well as others who need or prefer text; they are not simply an optional search-engine extra.

Put the transcript, or an obvious link to it, beside the video. Consider a London team sharing a training film with colleagues in Lagos: a readable text route also gives someone on a constrained connection a way to reach the instructions without streaming the clip. That is a practical publishing choice, not a claim about either city's network quality or a measured data saving.

Keep translation separate from accessibility checking

WAI distinguishes same-language captions from translated subtitles, while acknowledging that regions use the terms differently. A translation containing dialogue alone can still omit meaningful sounds. Start each language edition from checked source text, retain relevant labels, and have someone competent in that language review meaning and timing. A translated track is not, by itself, evidence that every accessibility need is covered.

Review the published experience, not just the files

Before release, check that the final page loads the intended caption track and transcript. Try the player without a mouse, look for visible keyboard focus, and check that controls have meaningful labels. These are among the features in WAI's media-player guidance, not a complete audit. Keep caption, narration and transcript revisions with the exact final video, so a later edit does not leave old text attached to a new cut.

Assign these jobs when commissioning the video, including outsourced work. The useful deliverable is a story people can follow through more than one route, not a folder containing an unchecked subtitle file. Follow PING's People desk for more on the design choices behind everyday technology.