Dec, 2020
贝叶斯 HMM 聚类 x 向量序列(VBx)在说话人分离中的理论、实现和标准任务分析
Bayesian HMM clustering of x-vector sequences (VBx) in speaker diarization: theory, implementation and analysis on standard tasks
Federico Landini, Ján Profant, Mireia Diez, Lukáš Burget
TL;DRVBx diarization method achieves superior performance on CALLHOME, AMI, and DIHARDII datasets by using a Bayesian hidden Markov model to find speaker clusters in x-vector sequences, and efficient formulae for the VBx model are presented while highlighting the lack of a standardized evaluation protocol for the AMI dataset in this paper.