Title :
Diarization resegmentation in the factor analysis subspace
Author :
Sell, Gregory ; Garcia-Romero, Daniel
Author_Institution :
Human Language Technol. Center of Excellence, Johns Hopkins Univ., Baltimore, MD, USA
Abstract :
Resegmentation is an important post-processing step to refine the rough boundaries of diarization systems that rely on segment clustering of an initial uniform segmentation. Past work has primarily used a Viterbi resegmentation with MFCC features for this purpose. In this paper, we examine an algorithm for resegmentation that operates instead in factor analysis subspace. By combining this system with a speaker clustering front-end, we yield a diarization error rate of 11.5% on the CALLHOME conversational telephone speech corpus.
Keywords :
speech recognition; CALLHOME conversational telephone speech corpus; MFCC; Viterbi resegmentation; diarization error rate; diarization resegmentation; diarization system; factor analysis subspace; rough boundaries; segment clustering; speaker clustering front-end; uniform segmentation; Clustering algorithms; Density estimation robust algorithm; Error analysis; Feature extraction; Hidden Markov models; Speech; Viterbi algorithm; Speaker diarization; factor analysis; variational Bayes;
Conference_Titel :
Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on
Conference_Location :
South Brisbane, QLD
DOI :
10.1109/ICASSP.2015.7178881
