When hearing music, it is natural for people to dance to its rhythm.
Automatic dance generation, however, is a challenging task due to the physical
constraints of human motion and rhythmic alignment with target music.
Conventional autoregressive methods introduce compounding errors dur
本研究提出了一种基于 Transformer 模型,结合之前姿势以及音乐情境来建模未来舞蹈动作分布的概率自回归模型,同时使用了包括专业舞者和业余舞者的当前最大的 3D 舞蹈动作数据集,通过物体评价和用户调查对比了两个基准模型,并表明要生成与音乐相匹配的有趣,多样和逼真的舞蹈,既需要模型具备建模概率分布的能力,又需要能够关注大范围的运动和音乐情境。