The speaking footage is done using Runway's Act One which allows human performance to drive an input image.
