Geometry Analysis and Recognition Research of Speech Signal
Xianbao Wang, Chunyan Xu, Chen Yong, Pan Xiaoxia, Wang Shoujue · 2008
Based on Principal Component Analysis technology, using geometry theory we analyze the distribution relations of single speech sample, samples of same class but of different people, samples of different classes of same people, and continuous speech samples of same people. Propose a covering recognition method of different classes of speech samples by complex geometrical bodies in feature space, and analyze the validity of it. Experiments show the rationality of this method.