Start with intent
Not every visible word is meant for the viewer. A lower-third name, warning, price, title card, or phone number is intentional communication. A distant road number, passing licence plate, newspaper page, or book spine may simply be part of the setting.
Ask one question: would a sighted viewer reasonably use this text to understand or act on the video? If yes, make it available. If no, do not crowd the soundtrack with it.
Text that usually needs description
- Opening and closing titles.
- Speaker names and roles.
- Instructions, warnings, dates, prices, URLs, phone numbers, and calls to action.
- Labels on a demonstrated control or object.
- Short chart labels or results that the speaker does not explain.
Text that usually does not
- Incidental signs that do not affect the story or task.
- Vehicle plates and background numbers.
- Long newspaper, textbook, or document pages used as scenery.
- Captions already carrying the spoken dialogue.
- Text that the speaker reads accurately at the same moment.
Handle animated text as one message
Many ads reveal a sentence word by word. Reading each fragment and then repeating the complete line wastes time. Wait until the statement is complete, then read it once at the closest usable window. If a changing word creates a deliberate contrast, describe that change instead.
When text appears over speech
Do not erase the speaker. If the text is essential and cannot wait, use an extended pause. If it is useful but not time-sensitive, place it in the next genuine gap. Keep phone numbers and URLs paced clearly enough to understand.
Keep names consistent
When a lower third identifies a person, carry that identity into later descriptions where the same person is clearly present. Do not guess that two similar-looking people are the same. Names can also be established by spoken self-identification, but the association should be supported by the video.
For a broader workflow, read How to Add Audio Description to a Video.