Articulatory-to-Acoustic Conversion of Mandarin Emotional Speech Based on PSO-LSSVM

Table 2

Comparison of RMSE in articulatory-to-acoustic conversion based on GMM, LSSVM, and PSO-LSSVM (Hz).

Acoustic feature	Emotion type	Number of samples in the test set	GMM	LSSVM	PSO-LSSVM
F2	Anger	29	32	28	23
	Happiness	32	31	26	22
	Neutral	28	43	35	26
	Sadness	23	42	33	31
	Average		36.52	30.20	25.10