Speaker adaptation in the maximum a posteriori framework based on the probabilistic 2-mode analysis of training models

EURASIP Journal on Audio, Speech, and Music Processing

Table 3 p -values from the matched-pair t -test

Methods	Number of adaptation sentences
	1	2	3	4	5
Prob. 2-mode and Tucker 3-mode	<0.01	0.22	0.08	0.03	0.34
Prob. 2-mode and MAPLR	0.10	< 0.01	< 0.01	0.01	0.02
Prob. 2-mode and MLLR, block-diagonal	<0.01	0.01	0.04	<0.01	0.05
Prob. 2-mode and EV	< 0.01
Tucker 3-mode and MLLR, block-diagonal	0.43	0.18	0.63	0.17	0.22
Tucker 3-mode and EV	0.94	< 0.01	< 0.01	< 0.01	< 0.01