Diffusion-based talking head models generate high-quality, photorealistic videos but suffer from slow inference, limiting practical applications. Existing acceleration methods for gen- eral diffusion models fail to exploit the temporal and spatial redundancies unique to talking head generation. In this paper, we propose a task-specific framework addressing these inefficiencies through two key innovations. First, we introduce Lightning-fast Caching-based Parallel denoising prediction (LightningCP), caching static features to bypass most model layers in inference time. We also enable parallel prediction using cached features and estimated noisy latents as inputs, efficiently bypassing sequential sampling. Second, we propose Decoupled Foreground Attention (DFA) to further accelerate attention computations, exploiting the spatial decoupling in talking head videos to restrict attention to dynamic foreground regions. Additionally, we remove reference features in certain layers to bring extra speedup. Extensive experiments demonstrate that our framework significantly improves inference speed while preserving video quality.

Emotional talking head generation has advanced significantly because of its potential to enhance naturalness and expressiveness in applications like gaming, virtual reality, and video conferencing. However, existing methods struggle with overcomplicated facial expression modeling, which introduces redundancies, reduces learning efficiency, and limits emotion accuracy. To address these limitations, we propose SimDEM, a novel framework for emotional talking head generation that leverages simplified and decoupled expression modeling. By retaining only critical dimensions of the expression representation via key expression modeling, we reduce noise in training and improve emotion accuracy. Furthermore, the decoupling of facial regions into independent dynamics for the upper and lower face enables more effective learning of localized emotional expressions, enhancing the visual fidelity of the output. Extensive evaluations demonstrate that SimDEM outperforms state-of-the-art methods, achieving superior emotion accuracy and visual quality in synthesizing expressive talking head videos.

With recent success of deep learning in 2-D visual recognition, deep-learning-based 3-D point cloud analysis has received increasing attention from the community, especially due to the rapid development of autonomous driving technologies. However, most existing methods directly learn point features in the spatial domain, leaving the local structures in the spectral domain poorly investigated. In this article, we introduce a new method, PointWavelet, to explore local graphs in the spectral domain via a learnable graph wavelet transform. Specifically, we first introduce the graph wavelet transform to form multiscale spectral graph convolution to learn effective local structural representations. To avoid the time-consuming spectral decomposition, we then devise a learnable graph wavelet transform, which significantly accelerates the overall training process. Extensive experiments on four popular point cloud datasets, ModelNet40, ScanObjectNN, ShapeNet-Part, and S3DIS, demonstrate the effectiveness of the proposed method on point cloud classification and segmentation.