The segmental bayesian information criterion and its applications to speaker diarization

Stafylakis, T; Katsouros, V; Carayannis, G

dc.contributor.author	Stafylakis, T	en
dc.contributor.author	Katsouros, V	en
dc.contributor.author	Carayannis, G	en
dc.date.accessioned	2014-03-01T01:34:47Z
dc.date.available	2014-03-01T01:34:47Z
dc.date.issued	2010	en
dc.identifier.issn	1932-4553	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/20853
dc.subject	Bayesian information criterion (BIC)	en
dc.subject	cluster analysis	en
dc.subject	clustering	en
dc.subject	speaker diarization (SD)	en
dc.subject.other	Asymptotic behaviors	en
dc.subject.other	Audio files	en
dc.subject.other	Bayesian information criterion	en
dc.subject.other	Cardinalities	en
dc.subject.other	clustering	en
dc.subject.other	Commonly used	en
dc.subject.other	Observation vectors	en
dc.subject.other	Operating points	en
dc.subject.other	Speaker diarization	en
dc.subject.other	Asymptotic analysis	en
dc.subject.other	Cluster analysis	en
dc.subject.other	Bayesian networks	en
dc.title	The segmental bayesian information criterion and its applications to speaker diarization	en
heal.type	journalArticle	en
heal.identifier.primary	10.1109/JSTSP.2010.2048656	en
heal.identifier.secondary	http://dx.doi.org/10.1109/JSTSP.2010.2048656	en
heal.identifier.secondary	5451053	en
heal.language	English	en
heal.publicationDate	2010	en
heal.abstract	This paper discusses the use of the BIC with respect to speaker diarization, i.e., the problem of assigning the observation vectors of an audio file to a set of speakers of unknown cardinality. Our primary goals are to examine the two dominant approaches of the BIC, namely the global and the local and combine the strengths of the two variants into one intuitive criterion, the segmental-BIC. We then consider the asymptotic behavior of the segmental-BIC, when dealing with models that are highly misspecified, as the ones commonly used in the speaker diarization task. Our main result is a modified version of the BIC, which significantly outperforms the current variants over the entire range of operating points, and achieves performance close to those of highly computationally demanding algorithms. © 2010 IEEE.	en
heal.publisher	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC	en
heal.journalName	IEEE Journal on Selected Topics in Signal Processing	en
dc.identifier.doi	10.1109/JSTSP.2010.2048656	en
dc.identifier.isi	ISI:000283266800010	en
dc.identifier.volume	4	en
dc.identifier.issue	5	en
dc.identifier.spage	857	en
dc.identifier.epage	866	en