Recently, He Shulin, a member of Professor Zhang Xueliang’s research group (IMUSPEECHLab) at our university, made significant progress in the field of Target Speaker Extraction (TSE). His research paper, titled “Enhancing Target Speaker Extraction with Hierarchical Speaker Representation Learning”, was published in Neural Networks, a flagship international journal in the fields of computer science and neural networks, with a current impact factor of 6.0.

Target Speaker Extraction, a promising alternative to conventional speech enhancement and separation techniques, has gained increasing attention in the field of intelligent audio processing. In this study, He Shulin proposed an innovative approach called Hierarchical Speaker Representation Learning (HSRL). Unlike traditional TSE methods, this work introduces a hierarchical strategy that integrates both local and global speaker feature extractors, substantially enhancing the performance of target speaker extraction systems.

This achievement marks another important milestone for IMUSPEECHLab in advancing cutting-edge research on intelligent speech processing technologies.