B-roll
The extra footage you lay over a talking head so viewers see what is being described instead of a face for four minutes.
See it
What it is
Your main footage (the interview, the piece to camera, the narration) is A-roll. B-roll is everything you cut over the top of it: hands typing, the wide shot of the building, the product spinning. The audio keeps running underneath, so the story never stops while the picture changes.
Two jobs, and the second one is the sneaky one. It shows what the voice is describing, and it hides edits: drop a cutaway over the seam and a jump cut disappears. That's why editors beg for more B-roll than a shoot ever plans for. Shoot the same action wide, medium, and tight, and hold each shot for a boring-feeling ten seconds so you have room to trim.
Gotcha: B-roll has to be motivated. Cut to the coffee cup while she's talking about coffee, not four sentences later. Random pretty shots read as filler and viewers feel the stall. Also let the cut breathe as an L-cut: bring the picture in a beat before or after the sentence lands, not exactly on it.
Ask AI for it
Edit this interview with B-roll coverage: keep the subject's audio running continuously as the A-roll spine, and cut supporting footage over the picture wherever the speaker names a concrete object, place, or action. Each B-roll shot runs 2 to 4 seconds, starts a half second before the matching word, and returns to the speaker for emphasis beats. Use cutaways to cover every jump cut in the A-roll so no edit is visible.