Hacker News new | ask | show | jobs
by jodrellblank 3023 days ago
Another one involving training a system on Obama speeches and lip movements.

Then being able to go from audio to synthesizing realistic mouth movements in a model and overlaying it on a video, to give a video of Obama "speaking" from cut-together audio, an impressionist, etc.

https://www.youtube.com/watch?v=9Yq67CjDqvw