Research on distribution of information of vowel in frequency domain by using sparse frequency components

In order to research relevant properties of speech signal, general analysis of speech signal sythesis adopts LPC parameters or formant frequency parameters to adjust them to get minimum error between natural speech and synthetic speech according to given poinciple of error. By using sparse components of speech spectrum to synthesize vowel through inverse Fonrier Transform, a few components of original spectrum are selected to synthesize vowel. The intelligibility of the synthesized vowels is listened respectively by a few listeners to examine its clarity and the distribution of information in speech signal in frequency domain. The result of the experiment is given and a conclusion is drawn.