Proposed was a method of smoothing the spectrum of vocal signal which is based on the histogram analysis of the peaks of the spectral envelope and represents a variety of the frequency analysis that is synchronous with the main tone. Its application was demonstrated by way of examples of the real vocal signals. It cannot be even doubted now that the linguistic information contained in the vocal signal is carried by the variations of the short-period amplitude spectrum of the signal reflecting both the method and place of sound generation and the prosodic speech characteristics. This fact is corroborated by the studies of speech production and peripheral auditory analyzer, various psychophysical experiments, and also the possibility of reading the spectrograms of the vocal signal.
展开▼