MOVA is an open-source AI that makes videos and sounds at the same time so mouths, actions, and noises match perfectly.
This paper shows a simple, one-model way to dub videos that makes the new voice and the lips move together naturally.