Table 1. Compounded feature set for classification.

feature		dimension
Spec	spectrum	128
	delta spectrum	128
Ceps	cepstrum	128
	delta cepstrum	128
AF	duration	1
	zero cross	1
	feature set *1	9
	delta feature set *2	9
	total	532

^*1.derived by non-linear transformations of spectrum and cepstrum: spectral centroid, spectral standard deviation, spectral skewness, spectral kurtosis, spectral entropy, spectral slope, peak quefrency, pitch goodness, and amplitude.

^*2.time derivatives of the nine features (*1).