Ignorant question ahead: how do the voices work with these things? A face, or entire body, is limited in its range of motion by our skeleton and muscles (for the large part. Puff your cheeks - there’s one of many exceptions). A voice, though, is MUCH more dynamic. Programming the lyrics and notes wouldn’t be nearly enough. Just getting the tone and inflections right seems like it would be an absolute nightmare.

I was wondering the same thing about that AI “actress”. In the case of a talented professional, (or even a hack who’s terrible but trying their best) a LOT of care and thought goes into the emotion behind each word. How do they program that? Or are these just fancy 3D puppets with human voice actors behind them?