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

Table 5

Average emotion recognition rates for SES database.

Different order “db” wavelets	Speaker dependent
	Raw features	Enhanced features	SET1	SET2	SET3	SET4
ELM kernel
db3	41.93	89.92	90.38	89.00	88.46	86.21
db6	42.14	92.79	91.21	92.04	90.63	91.21
db10	40.90	88.83	89.96	90.04	88.58	88.67
db44	42.38	90.00	89.79	88.83	84.67	82.13
kNN
db3	31.15	77.79	79.50	78.00	78.42	76.88
db6	32.80	81.79	82.17	82.96	81.46	81.63
db10	31.23	78.63	79.38	80.08	79.79	81.08
db44	32.08	78.79	79.38	78.25	74.63	73.71
Speaker independent
ELM kernel
db3	27.25	78.75	79.17	78.25	78.50	77.50
db6	26.00	83.67	83.75	84.58	83.58	83.42
db10	27.08	79.42	80.42	80.25	80.33	79.92
db44	26.00	80.50	80.92	78.67	76.33	74.92
kNN
db3	25.92	69.33	69.67	67.92	68.17	66.83
db6	25.50	71.92	73.67	74.00	72.17	73.50
db10	26.33	70.00	71.00	71.08	71.17	70.92
db44	24.67	67.58	69.50	68.08	66.08	67.83