Speaker Diarization with Region Proposal Network
Abstract
Speaker diarization is an important pre-processing step for many speech applications, and it aims to solve the "who spoke when" problem. Although the standard diarization systems can achieve satisfactory results in various scenarios, they are composed of several independently-optimized modules and cannot deal with the overlapped speech. In this paper, we propose a novel speaker diarization method: Region Proposal Network based Speaker Diarization (RPNSD). In this method, a neural network generates overlapped speech segment proposals, and compute their speaker embeddings at the same time. Compared with standard diarization systems, RPNSD has a shorter pipeline and can handle the overlapped speech. Experimental results on three diarization datasets reveal that RPNSD achieves remarkable improvements over the state-of-the-art x-vector baseline.
Cite
@article{arxiv.2002.06220,
title = {Speaker Diarization with Region Proposal Network},
author = {Zili Huang and Shinji Watanabe and Yusuke Fujita and Paola Garcia and Yiwen Shao and Daniel Povey and Sanjeev Khudanpur},
journal= {arXiv preprint arXiv:2002.06220},
year = {2020}
}
Comments
Accepted to ICASSP 2020