New audio-visual fusion model powers lifelike AI humans by interpreting tone, facial expression, posture, and speech nuances—now generally available across Tavus conversations and APIs.