Research Article

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

Table 3

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

Acoustic featureEmotion typeNumber of samples in the test setGMMLSSVMPSO-LSSVM

MFCCAnger292.211.981.44
Happiness322.121.691.49
Neutral282.462.031.50
Sadness232.881.961.63
Average2.3841.8251.508