未提供具体数据集名称
收藏资源简介:
该数据集由浦项科技大学的研究团队创建,用于训练和评估3D说话人头部的生成模型。数据集通过大量的2D同步说话脸视频学习得到音频-视觉语音表示,并进一步与3D面部网格结合,构建了丰富的语音-网格同步表示空间。该空间能够捕捉到语音和唇部运动之间的复杂对应关系,用于提高现有3D说话人头模型中唇部同步的感知准确性。
This dataset was developed by a research team at Pohang University of Science and Technology for training and evaluating generative models for 3D talking head generation. It extracts audio-visual speech representations from a large-scale corpus of synchronized 2D talking-face videos, and further integrates these representations with 3D facial meshes to construct a rich synchronized speech-mesh representation space. This space captures the complex correspondences between speech and lip movements, aiming to improve the perceptual accuracy of lip synchronization in existing 3D talking head models.




