Latent class model with application to speaker diarization

EURASIP Journal on Audio, Speech, and Music Processing

Table 3 Experiment results of different methods on RT09

DER[%]	Speaker #	BIC	VB	LCM-Ivec
				PLDA	SVM	Hybrid
Given speaker #	-	Yes	Yes	Yes	Yes	Yes
EDI_20071128-1000	4	29.32	10.67	9.89	9.91	9.83
EDI_20071128-1500	4	35.61	48.66	19.68	19.87	17.40
IDI_20090128-1600	4	29.12	11.15	7.02	7.14	7.14
IDI_20090129-1000	4	37.27	35.85	31.99	32.37	21.82
NIST_20080201-1405	5	61.54	49.05	44.67	43.05	38.53
NIST_20080227-1501	6	40.32	39.97	24.76	25.66	13.96
NIST_20080307-0955	11	46.62	23.50	22.86	16.44	16.00
Average	-	39.97	31.26	22.98	22.06	17.81

¹The code for the BIC diarization system was downloaded from: https://github.com/gdebayan/Diarization_BIC
²VB is the system described in P. Kenny’s paper [2]. This system is partly realized by the python code downloaded from: http://speech.fit.vutbr.cz/software/vb-diarization-eigenvoice-and-hmm-priors