Researcher Harry McGurk learned the answer by accident. A technician gave a video a new soundtrack, so the voice and lips made different sounds. McGurk and his assistant heard a third sound.
Play a voice saying “ba-ba” over lips saying “ga-ga,” and people hear “da-da.”
The brain merges sight and sound automatically. Knowing the trick doesn't stop it: some researchers who've studied it for over 20 years still hear it.