Some of these neural network things can track the speaker's face. So not necessary, but it might might make things more interesting.