Comparison of ALBAYZIN query-by-example spoken term detection 2012 and 2014 evaluations

EURASIP Journal on Audio, Speech, and Music Processing

Table 11 Results of the ALBAYZIN QbE STD 2014 evaluation on foreign queries of the test data

System ID	MTWV	ATWV	p(FA)	p(Miss)
Fusion (SGMM) + Posteriorgram	0.5167	0.5167	0	0.483
Fusion (SGMM)	0.5167	0.5167	0	0.483
Fusion (DNN) + Posteriorgram	0.5500	0.5500	0	0.450
Fusion (DNN)	0.5500	0.5500	0	0.450
LVCSR-SGMM	0.2667	0.2667	0	0.733
LVCSR-DNN	0.3000	0.3000	0	0.700
MFCC-Fingerprint	0.3833	0.3833	0	0.617
Phoneme Posteriorgram	0.3833	0.3833	0	0.617
Text-based SGMM STD	0.3675	0.3400	0.00008	0.550
Text-based DNN STD	0.4225	0.4225	0.00003	0.500