Improved Emotion Recognition Using Gaussian Mixture Model and Extreme Learning Machine in Speech and Glottal Signals

Table 2

Number of selected enhanced features using SWLDA.

Different order “db” wavelets	value	BES				SAVEE				SES
		Features from speech signals		Features from glottal signals		Features from speech signals		Features from glottal signals		Features from speech signals		Features from glottal signals
		Number of selected RWPEGYs	Number of selected RWPEPYs	Number of selected RWPEGYs	Number of selected RWPEPYs	Number of selected RWPEGYs	Number of selected RWPEPYs	Number of selected RWPEPYs	Number of selected RWPEPYs	Number of selected RWPEGYs	Number of selected RWPEPYs	Number of selected RWPEPYs	Number of selected RWPEPYs
db3	0.05–0.1	21	26	12	12	18	21	10	13	19	25	16	19
	0.01–0.05	17	22	8	11	17	20	7	10	16	24	13	16
	0.001–0.01	16	18	7	8	14	17	6	5	15	22	5	8
	0.0001–0.001	7	13	5	7	13	13	5	5	8	18	5	6
db6	0.05–0.1	18	23	13	14	18	20	17	19	18	23	13	14
	0.01–0.05	13	22	7	12	16	21	17	17	13	22	7	12
	0.001–0.01	12	19	7	10	13	19	11	10	12	19	7	10
	0.0001–0.001	14	13	7	7	13	19	11	10	14	13	7	7
db10	0.05–0.1	22	22	12	12	5	9	4	4	20	23	12	13
	0.01–0.05	17	17	11	10	5	9	4	4	19	19	12	12
	0.001–0.01	14	15	6	7	5	9	4	4	16	18	9	9
	0.0001–0.001	14	15	6	5	5	9	4	3	12	11	6	8
db44	0.05–0.1	25	27	14	15	14	20	7	11	25	27	14	15
	0.01–0.05	20	24	9	9	14	20	7	10	20	24	9	9
	0.001–0.01	15	15	8	7	14	19	6	9	15	15	8	7
	0.0001–0.001	12	13	4	5	13	18	5	3	12	13	4	5