A video announcement should ship with accurate captions, a useful transcript and a plan for important visual information. Build those assets while the video is produced, then verify them in the public player instead of treating accessibility as a post-publication note.
Captions, transcripts and audio description do different jobs. A supplier’s package should name each deliverable separately, including the language, file format, review process and public placement.
Separate the deliverables
Captions are synchronized text for speech and non-speech audio needed to understand a video. A transcript presents that information as readable page text. A descriptive transcript also adds visual information that matters when someone cannot see the video.
Audio description communicates important visual information for people who are blind or have low vision. It can be integrated into the main script, supplied as a separate track or provided in another supported format.
Use a scope table before production:
| Asset | Minimum scope to agree | Public check |
|---|---|---|
| Captions | Language, speakers, meaningful sounds, timing and file format | Turn captions on and watch the full video |
| Transcript | Speakers, speech, important sounds, headings and links | Open the transcript beside the public player |
| Visual description | On-screen text, charts, actions and other essential visuals | Review without relying on sight |
| Player | Keyboard access, caption control and supported tracks | Test the actual embedded player |
Do not call translated subtitles a replacement for same-language captions. W3C uses captions for the same language as the audio and subtitles for translations.
Write accessibility into the script
For a new product video, let speakers say the important visual facts where that sounds natural. Instead of “as you can see here,” name the chart’s change, the control being selected or the status shown on screen.
W3C notes that integrated description can reduce later production work for suitable instructional videos. It also improves the ordinary script because the demonstration is understandable without unexplained gestures.
Plan pauses for visuals that need separate description. If a presenter talks continuously while several screens change, there may be no room for an additional description track without editing.
Edit automatic captions
Automatic captions are a starting point, not acceptance evidence. W3C says they do not meet user needs or accessibility requirements unless they are confirmed to be fully accurate.
Check names, product terms, numbers, negations and market restrictions. One missing “not” can reverse a claim. Include speaker identification and meaningful non-speech audio such as an alert tone when it affects understanding.
WebVTT is the most common web caption format; SRT and TTML are also used. Agree the required format with the publisher and retain the editable source alongside the final export.
Build the transcript from reviewed text
If the captions are approved, reuse their text for the transcript. Remove timing clutter unless timestamps help navigation, combine short caption lines into paragraphs and add headings and relevant links.
Review the video again for visual information that captions may omit: speaker names, on-screen URLs, chart values, error messages and steps performed without narration. W3C recommends adding those details to a descriptive transcript.
Put the transcript or a clear transcript link close to the video. On a company announcement page, the same-page option also gives readers a durable text record when playback is inconvenient.
Review the public player
Test with a keyboard and inspect the controls at desktop and mobile sizes. Confirm that captions can be turned on, the correct language is selected, text is not clipped and the transcript link remains visible.
Some media players can turn a caption file into an interactive transcript, highlighting phrases as they are spoken and letting a reader jump to a point in the video. Treat that feature as a player capability, not a property of the caption file alone.
Check the thumbnail, surrounding copy and destination links too. The crypto creator brief covers disclosure, claims and reuse rights. The image accessibility checklist covers still images and chart context.
Accept the delivered package
Keep the approved script, final video, caption source, transcript, description plan and public URL together. Record the language and version of each asset.
Watch the public version from start to finish with captions enabled. Then read the transcript without playing the video. Correct missing or conflicting facts before the campaign is reported as delivered.
A production file in a shared folder proves that an asset exists. It does not prove that the public page uses the right version. Publication evidence and performance measurement remain separate checks.
Sources and review date
Sources checked October 23, 2026. Requirements vary by media, player, audience and applicable standard; this guide is an editorial production checklist, not legal advice.