Every month, the joint laboratory invites external speakers to take part in seminars for its partners.
Shuai Wang (School of Intelligence Science and Technology at Nanjing University): « Towards Real-world Target Speech Extraction: Algorithms, Dataset and Toolkits«
Abstract: Target speaker extraction has attracted increasing attention from researchers due to its significant application potential. Despite substantial progress in this field, a considerable gap remains between current research and real-world scenarios. In this presentation, I will first discuss our algorithmic efforts to enhance the robustness of TSE systems, then introduce the datasets we have developed for more realistic evaluation of real-world applications, as well as WeSep, an open-source toolkit specifically designed for target speaker extraction tasks.
Bio:
Dr. Shuai Wang is currently an Associate Professor in the School of Intelligence Science and Technology at Nanjing University. His research encompasses speaker modeling, target speaker processing, speech generation and music generation. He was the winner of international challenges such as VoxSRC 2019 and DIHARD 2019, and was honored with the IEEE Ganesh N. Ramaswamy Memorial Grant Award in 2019. He received both the Best Paper and Best Student Paper Awards at ISCSLP 2024. Dr. Wang initiated the open-source WeSpeaker and WeSep toolkits, with the pre-trained models achieving millions of downloads monthly on Hugging Face and adopted by researchers across academic research and industrial applications.
