Xiao Sha

Hebeizi Li, Zihao Liang, Benyuan Sun et al.

While state-of-the-art audio-video generation models like Veo3 and Sora2 demonstrate remarkable capabilities, their closed-source nature makes their architectures and training paradigms inaccessible. To bridge this gap in accessibility and performance, we introduce UniTalking, a unified, end-to-end diffusion framework for generating high-fidelity speech and lip-synchronized video. At its core, our framework employs Multi-Modal Transformer Blocks to explicitly model the fine-grained temporal correspondence between audio and video latent tokens via a shared self-attention mechanism. By leveraging powerful priors from a pre-trained video generation model, our framework ensures state-of-the-art visual fidelity while enabling efficient training. Furthermore, UniTalking incorporates a personalized voice cloning capability, allowing the generation of speech in a target style from a brief audio reference. Qualitative and quantitative results demonstrate that our method produces highly realistic talking portraits, achieving superior performance over existing open-source approaches in lip-sync accuracy, audio naturalness, and overall perceptual quality.

Xiao Sha, Zhu Sun, Jie Zhang

Knowledge graphs (KGs) have proven to be effective for high-quality recommendation, where the connectivities between users and items provide rich and complementary information to user-item interactions. Most existing methods, however, are insufficient to exploit the KGs for capturing user preferences, as they either represent the user-item connectivities via paths with limited expressiveness or implicitly model them by propagating information over the entire KG with inevitable noise. In this paper, we design a novel hierarchical attentive knowledge graph embedding (HAKG) framework to exploit the KGs for effective recommendation. Specifically, HAKG first extracts the expressive subgraphs that link user-item pairs to characterize their connectivities, which accommodate both the semantics and topology of KGs. The subgraphs are then encoded via a hierarchical attentive subgraph encoding to generate effective subgraph embeddings for enhanced user preference prediction. Extensive experiments show the superiority of HAKG against state-of-the-art recommendation methods, as well as its potential in alleviating the data sparsity issue.