Listen before you describe
Play the video without looking at the picture. Note every moment where a listener cannot answer a basic question that a sighted viewer can: Who entered? What changed? What does the title card say? Which speaker is on screen? What step is being demonstrated?
Do not write narration for information the dialogue already communicates. Repetition makes the track tiring and takes space away from details that matter.
Build a visual information list
Review the picture and record only information that changes understanding:
- Names, roles, locations, dates, prices, phone numbers, and calls to action shown intentionally on screen.
- Actions whose result is not clear from sound alone.
- Scene changes needed to follow the story or task.
- Speaker identity when the voice cannot establish it.
- Charts, demonstrations, gestures, and visual comparisons that carry meaning.
Ignore decorative textures, incidental licence plates, distant signage, and pages of background print unless the video directs the viewer to them.
Choose standard or extended description
Standard audio description places narration inside available gaps. It works well when music, silence, or non-essential sound leaves enough time. Extended audio description pauses the picture and program audio when essential information cannot fit safely.
The W3C G8 technique describes pausing a movie when critical visual information needs more room. Use a pause because the information matters, not because the script is wordy.
Write for the ear
Use active language: “Maya signs the form” is clearer than “The scene shows a visual depiction of Maya signing.” Name the subject, action, and useful object. Put the most important fact first because the available window may be short.
Read purposeful text exactly when exact wording matters. For a long slide, convey the information needed to understand the video rather than reading every word. If text appears one phrase at a time, wait until the complete statement is visible unless an earlier fragment changes the meaning.
Place narration without losing the program
- Prefer silence and genuine music windows.
- Lower music while narration speaks so both remain intelligible.
- Do not cover dialogue for ordinary scene detail.
- If essential visible text appears over speech, add a synchronized pause rather than erasing the speaker.
- Leave enough separation between descriptions for the program to breathe.
Review the published experience
Check the described video from beginning to end. Confirm that names stay consistent, text is not repeated, the original speech remains complete, pauses resume at the correct frame, and the AD control works by keyboard and screen reader. Captions need their own review on the original timeline.
VoxDescribe creates all four working files and lets you edit AD and captions before rebuilding the downloads. See the audio description generator or process a five-minute test video free.