flâneur

Modeling natural conversational dynamics with Seamless Interaction

ai.meta.com · 1,845 words · saved by 1 readers

Communication between people is like a dance, with each person continuously adjusting what they say, how they say it, and how they gesture. Modeling two-party, or dyadic, conversation dynamics entails understanding the multimodal relationship between vocal, verbal, and visual social signals—and the interpersonal behaviors between people, such as listening, visual synchrony, and turn-taking. As virtual agents become important helpers in our daily lives, it's important that these systems be able to display these natural patterns of conversation. Today, the Meta Fundamental AI Research (FAIR) team together with Meta’s Codec Avatars lab and Core AI lab are introducing a family of Dyadic Motion Models that explore new frontiers of social AI. These models render human or language model generated speech between two individuals into diverse, expressive full-body gestures and active listening behaviors, allowing the creation of fully embodied avatars in 2D video and as 3D Codec Avatars. The mod

Modeling natural conversational dynamics with Seamless Interaction Products AI Research Resources About Meta Model API Try Meta AI Open Source Modeling natural conversational dynamics with Seamless Interaction June 27, 2025 • 12 minute read Takeaways: As we work to build the future of human connection and the technology that makes it possible, we’ll need models that can generate facial expressions and body gestures based on audio-visual inputs from two people. Meta Fundamental AI Research (FAIR) is introducing a family of audiovisual behavioral motion models to address that need. Our models al

related reading