1. Treat the approved narration as the source of truth
Draft captions from the audio you will actually publish, not from an earlier script. Small changes to names, numbers, pauses, or pronunciation can make a transcript inaccurate even when the original copy looked correct.
Keep a reviewed transcript next to the audio export. It helps editors, translators, support teams, and people who cannot listen to understand the same message without guessing what changed.
2. Design captions for reading, not just transcription
Use short lines, natural phrase breaks, and enough time for a viewer to finish reading. Do not hide an important product name under a button, face, or busy lower-third. Preview on a phone rather than trusting a desktop timeline.
Check contrast against every shot. If the background changes quickly, use a consistent caption treatment that remains legible without turning the video into a wall of decoration.
- Keep a speaker change or idea change visible in the line break.
- Spell product names, numbers, and URLs exactly as approved.
- Review timing after the last audio edit, not before.
3. Put visual information into words when it matters
If a screen recording shows a result that the narration never names, a listener may miss the point. Add a short spoken explanation, a text alternative, or a transcript note when the visual is essential to the task.
Audio description is not a license to narrate every decorative movement. Prioritize information that changes what a person can understand or do. Keep the wording concise and test it with someone who was not involved in the edit.
4. Test the finished package
- Listen once with the screen off and confirm the main idea still makes sense.
- Watch once with sound off and confirm the captions carry the essential message.
- Check captions on a small phone at ordinary brightness and volume.
- Ask at least one person outside the edit to flag confusing terms or timing.
- Keep the transcript and final audio version together for future updates.
