Recognition of Emotions in Mexican Spanish Speech: An Approach Based on Acoustic Modelling of Emotion-Specific Vowels

Table 9

Average classification performance of the ASR system for speech and emotion recognition (individual speakers from the speech corpus) across five iterations.

Speaker	Statistic	Anger	Happiness	Neutral	Sadness	Word accuracy	Phoneme accuracy
MS1	Average	100.00	66.67	100.00	100.00	78.75	82.82
	Std Dev	0.00	0.00	0.00	0.00	2.17	1.71
MS2	Average	81.90	86.67	100.00	100.00	86.59	88.32
	Std Dev	11.74	12.64	0.00	0.00	5.21	4.70
MS3	Average	90.00	95.00	100.00	100.00	93.17	93.83
	Std Dev	9.13	11.18	0.00	0.00	3.52	3.09
FS1	Average	96.67	100.00	93.33	100.00	77.00	82.34
	Std Dev	7.45	0.00	14.91	0.00	6.59	4.41
FS2	Average	76.31	100.00	100.00	100.00	83.63	85.88
	Std Dev	10.85	0.00	0.00	0.00	5.61	4.96
FS3	Average	77.26	100.00	100.00	100.00	84.75	86.46
	Std Dev	9.96	0.00	0.00	0.00	7.56	5.85
Total (all speakers)	Average	87.02	91.39	98.89	100.00	83.98	86.61
	Std Dev	12.47	13.75	6.09	0.00	7.30	5.55