EURASIP Journal on Audio, Speech, and Music Processing

Table 3 Confusion matrix for HMM-based system using formant parameter contours (S-I), $ν^{(i)} = [s_{F_{1}}^{(i)}, b_{F_{1}}^{(i)}, s_{F_{2}}^{(i)}, b_{F_{2}}^{(i)}, s_{F_{3}}^{(i)}, b_{F_{3}}^{(i)}, s_{A_{1}}^{(i)}, b_{A_{1}}^{(i)}, s_{A_{2}}^{(i)}, b_{A_{2}}^{(i)}, s_{A_{3}}^{(i)}, b_{A_{3}}^{(i)}]$ (overall accuracy 41.6%)

From: On the use of speech parameter contours for emotion recognition

	Neutral (%)	Anger (%)	Sad (%)	Happy (%)	Bored (%)
Neutral	55.3	2.1	14.9	17.0	10.6
Anger	4.2	46.5	0	26.8	9.9
Sad	6.6	8.2	32.8	24.6	27.9
Happy	1.4	21.9	15.1	37.0	24.7
Bored	2.6	9.1	20.8	27.3	40.3

Back to article page