Cost-Sensitive Learning for Emotion Robust Speaker Recognition

Table 1

Comparison of system performance under different types of affective speech (%).

Method	Baseline	CSSR
Anger	21.80	33.74
Elation	22.70	36.23
Neutral	94.40	95.63
Panic	26.30	36.14
Sadness	51.13	54.67
Total	43.27	51.28