Meta AI Twin: A Smart Virtual Meeting Platform with 2D Avatar-Based User Presence and Cloud-Driven Language Services

This paper aims to design and demonstrate an online platform where members can join a group with a meeting code. Inside, the participant can have real-time conversations with a 2D avatar. The minutes of the meeting are logged with the help of Microsoft’s Speech-to-Text, and people can join at the appointed time and even set their 2D avatars to unavailable, signaling the end of the meeting when everyone is offline. In other words, the end-product increases usability, productivity, and accessibility. Thus, this technology is poised for future enhancements in various ways: multilingual chat, interpretation, and summaries; more lifelike and personalized 2D avatars of each participant that are more real; and AI-powered decision support tools to track decisions, action items, and questions using OpenAI and similar provider’s language-model technology. This could be extended to audio and video summarization, especially when shared screens are utilized in presentations, and speech summarization under noisy environments. However, security and authentication related issues would be the molders in how we will develop a truly smart and virtual meeting area.

[1]    K. Huang, Y. Li, and S. Liu, ”DreamWaltz: Scene-level Animatable Avatars from Internet Videos,” in Proceedings of the IEEE/CVF Con-ference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 1234-1245. [2]    A. Aliakbarian, M. R. Osmani, and H. Escalante, ”HMD-NeMo: Online 3D Avatar Motion Generation from Sparse Observations,” ACM Transactions on Graphics (TOG), vol. 43, no. 2, pp. 1-13, 2024. [3]     L. Zhang, J. Wu, and T. Chen, ”MagicAvatar: Multimodal Avatar Gener-ation and Animation for Interactive Communication,” IEEE Transactions on Visualization and Computer Graphics, vol. 30, no. 5, pp. 2678-2690, 2024. [4]     K. Huang, X. Zhao, and L. Wang, ”WildAvatar: Learning 3D Avatars In-the-Wild from Large-Scale Internet Data,” in European Conference on Computer Vision (ECCV), 2024, pp. 543-559. [5]     Y. Wu, Q. Li, and M. J. Black, ”AnimPortrait3D: Text-Based Animatable Avatars via Morphable Model Alignment,” Computer Graphics Forum, vol. 43, no. 7, pp. 1122-1135, 2024. [6]     S. Yoshida, Y. Koyama, and Y. Ushiku, ”Towards AI-Mediated Avatar-Based Telecommunication: Investigating Visual Impression of Switching Between Avatars,” in Proceedings of the ACM Symposium on Virtual Reality Software and Technology, 2024. [7]     A. Zainudin, ”Collaborative Edge Learning-Assisted Interactive Avatar Integration with Human Activity Recognition,” Engineering Reports, vol. 7, no. 2, pp. 45-62, 2025. [8]    R. Zhu, ”Avatar Design in Metaverse: The Effect of Avatar-User Similarity on Task Engagement,” Computers in Human Behavior, vol. 139, pp. 107-120, 2023. [9]    G. C. Dobre, ”Avatars in Mixed-Reality Meetings: Effects on Com-munication, Presence, and Mood Perception,” University of Augsburg Technical Report, no. TR-2025-01, 2025. [10]    C. Gasch, ”A Study on User Acceptance, Preferences, and Technostress in Virtual Meetings with Avatars,” Applied Sciences, vol. 15, no. 6, pp. 2345-2360, 2025.