Source Separation
Source separation aims to decompose complex audio mixtures into their constituent sources, such as individual instruments in a musical recording or speakers in a conversation. Current research emphasizes unsupervised and weakly-supervised approaches, employing diverse architectures including diffusion models, variational autoencoders, and neural networks combined with traditional signal processing techniques like non-negative matrix factorization and beamforming. These advancements are improving the robustness and efficiency of source separation across various applications, from enhancing speech recognition in noisy environments to enabling more sophisticated music generation and analysis.
92papers
Papers
January 27, 2025
December 9, 2024
October 31, 2024
October 29, 2024
September 18, 2024
September 17, 2024
September 3, 2024
The USTC-NERCSLIP Systems for the CHiME-8 NOTSOFAR-1 Challenge
Shutong Niu, Ruoyu Wang, Jun Du, Gaobin Yang, Yanhui Tu, Siyuan Wu, Shuangqing Qian, Huaxin Wu, Haitao Xu, Xueyang Zhang, Guolong Zhong+9Activity-Guided Industrial Anomalous Sound Detection against Interferences
Yunjoo Lee, Jaechang Kim, Jungseul Ok
August 19, 2024
August 12, 2024
June 30, 2024