EURASIP Journal on Audio, Speech, and Music Processing

Table 3 Results for valence-based classification and regression on SEMAINE

From: Within and cross-corpus speech emotion recognition using latent topic model-based features

Metric	IS09	VQ	LDA	sLDA	RSM	sRSM
	Development set
WA	59.61	58.33	63.03	64.10	65.50	66.45
UA	57.64	56.51	58.86	61.96 ^†	63.53 ^∗†	64.62 ^∗†
COR	0.191	0.191	0.330	0.332	0.327	0.349
	Test set
WA	49.93	51.56	55.13	57.21	56.32	57.80
UA	49.68	51.54	56.12 ^∗	57.63 ^∗†	56.88 ^∗†	58.00 ^∗†
COR	0.007	0.045	0.128	0.154	0.127	0.171

Classification results are expressed in percentage (%). Symbols ∗ and † indicate statistical significance over IS09 and VQ, respectively.

Back to article page