Avatar Forcing：面向自然对话的实时交互式头部虚拟人生成
Avatar Forcing: Real-Time Interactive Head Avatar Generation for Natural Conversation
扩散强制（diffusion forcing）框架将自回归生成与扩散模型结合，实现因果约束下的实时多模态条件生成；直接偏好优化（DPO）利用合成负样本进行无标签的偏好学习，可迁移到其他需要实时交互的生成任务。
原文依据 we propose Avatar Forcing, a new framework for interactive head avatar generation that models real-time user-avatar interactions through diffusion forcing.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论说话头生成技术能够从静态肖像创建逼真的虚拟人，用于虚拟通信和内容创作。然而，当前模型尚未传达出真正交互式通信的感觉，往往生成缺乏情感投入的单向响应。我们指出了实现真正交互式虚拟人的两个关键挑战：在因果约束下实时生成运动，以及在没有额外标注数据的情况下学习富有表现力且生动的反应。为解决这些挑战，我们提出了 Avatar Forcing，一种通过扩散强制（diffusion forcing）来建模实时用户-虚拟人交互的交互式头部虚拟人生成新框架。该设计允许虚拟人处理实时多模态输入，包括用户的音频和运动，并以低延迟对言语和非言语线索（如语音、点头和笑声）做出即时反应。此外，我们引入了一种直接偏好优化方法，利用通过丢弃用户条件构建的合成负样本，实现无需标签的表达性交互学习。实验结果表明，我们的框架能够以低延迟（约500毫秒）实现实时交互，相比基线获得6.8倍加速，并产生具有反应性和表现力的虚拟人运动，在超过80%的情况下优于基线。
英文原文摘要 1344 字符
Talking head generation creates lifelike avatars from static portraits for virtual communication and content creation. However, current models do not yet convey the feeling of truly interactive communication, often generating one-way responses that lack emotional engagement. We identify two key challenges toward truly interactive avatars: generating motion in real-time under causal constraints and learning expressive, vibrant reactions without additional labeled data. To address these challenges, we propose Avatar Forcing, a new framework for interactive head avatar generation that models real-time user-avatar interactions through diffusion forcing. This design allows the avatar to process real-time multimodal inputs, including the user's audio and motion, with low latency for instant reactions to both verbal and non-verbal cues such as speech, nods, and laughter. Furthermore, we introduce a direct preference optimization method that leverages synthetic losing samples constructed by dropping user conditions, enabling label-free learning of expressive interaction. Experimental results demonstrate that our framework enables real-time interaction with low latency (about 500ms), achieving 6.8x speedup compared to the baseline, and produces reactive and expressive avatar motion, which is preferred over 80% against the baseline.