SRIBD Seminar:Deep speaker embedding learning and applications to related tasks by Dr. Shuai Wang(Online)
Speaker: Dr. Shuai Wang
Topic: Deep speaker embedding learning and applications to related tasks
Time & Date: 10:30 -11:30, Tuesday,January 17, 2023 (Beijing time)
Zoom Meeting ID:940 4518 6370
Speaker embeddings are low-dimensional representations of a speaker's voice that capture his unique characteristics. They are commonly used for tasks where speaker identity should be modeled. In this talk, I will introduce several approaches for improving speaker embedding's robustness and how to use the embeddings in tasks such as speaker recognition, speech synthesis, and voice conversion.
Shuai Wang obtained his Ph.D. degree at Shanghai Jiao Tong University in 2020.09, under the supervision of Kai Yu. During his Ph.D., he had been working on speaker identity modeling, he had published more than 30 related papers at top-tier conferences and journals in the speech processing area, with more than 1000 citations. Shuai Wang was the winner of several international competitions such as VoxSRC2019 and DIHARD 2019. After graduation, he joined Tencent as a senior research scientist, working on application-oriented speech-processing algorithms. Shuai is also a member of wenet open-source community, he initiated the wespeaker project, which provides a SOTA speaker embedding learning framework and has been adopted by researchers from the academia and industry.