DiffTalk:制作广义音频驱动人像动画的扩散模型paper：DiffTalk:CraftingDiffusionModelsforGeneralizedAudio-DrivenPortraitsAnimation(thecvf.com)code：GitHub-sstzal/DiffTalk:[CVPR2023]Theimplementationfor"DiffTalk:CraftingDiffusionModelsforGeneralizedAudio-DrivenPortraitsAnimation"目录1介绍2背景3方法4实验5结论1介绍生成质量和模型泛化是将谈话头合成技术的两个重要因