Lip sync
Lip sync is how closely an AI Avatar's mouth movements match the words it speaks. Good lip sync is frame-accurate and natural; poor lip sync, even by a fraction of a second, is the fastest way an Avatar stops feeling human.
How AI lip sync works
In an audio-driven avatar, the model reads the incoming speech and generates mouth shapes, facial movement and head motion that match it frame by frame. Because it follows the audio rather than the text, it works for any language the voice can speak.
In real time, lip sync has to be produced as fast as the audio streams in, and stay in sync even on unstable networks.
How Beyond Presence does it
Genesis 2.0 is audio-driven, with precise lip sync, natural head motion and empathetic listening animations, streamed at 1080p and 35 FPS. In a blind study run by an independent researcher, 119 participants rated Genesis 2.0 highest on visual quality of four real-time providers. Read the study.
Frequently asked questions
How does Beyond Presence handle lip sync?
Genesis 2.0 renders the face from the audio itself, using a 3D model trained for each person, so lips, expressions and head motion follow every word in about 100 ms.
Does Beyond Presence lip sync work in every language?
Yes. It follows the audio, so it works in any language the voice speaks: 30+ with Managed Agents and 100+ with the Speech-to-Video API.
How good is Beyond Presence's visual quality?
In a blind study with 119 participants, Genesis 2.0 was rated highest on visual quality of four providers. Talk to Laura, our customer support Avatar, to judge it yourself.