Where captions serve deaf and hard-of-hearing viewers by providing text for audio, audio description serves blind and low-vision viewers by providing spoken description for visuals that the existing dialogue does not already convey, such as a character entering a room silently or an important piece of on-screen text.
WCAG requires audio description for prerecorded video at Level AA (success criterion 1.2.5), and it matters most for content where meaning depends heavily on visuals the dialogue does not explain, such as instructional or demonstration videos.