1 paper
Artem Kravtsov, Dmitrii Ziganshin, Vsevolod Poletaev +4
Talking-head and dyadic models now achieve real-time inference, yet fast motion generation alone does not produce an interactive conversation. A live system must synchronize the mo…