METHOD FOR CEPSTRAL IDENTIFICATION OF SIGNALS

摘要

The invention relates to means for automated cepstral identification, selection and identification of sources of language hydro-acoustic, bio-medical signals; signals of seismic origin, ultrasonic signals. Method forc identification of signals is in fact that one performs low-frequency filtration of signal in band of frequencies of analysis, signal is separated to segments of analysis with same duration agreed with interval of correlation of signal, one forms on each segment of analysis values of discrete time counts of signal through use of standard analog-digital transformation with discretization frequency by time agreed with band of frequencies of spectrum analysis, by quantization pitch, by level agreed with dynamical range of amplitudes, on each segment of analysis one forms discrete counts of current spectral density of power of signal through use of standard discrete Fourier transform, one performs standard logarithmic transformation of valus of current spectral density of power of signal by all the segments of analysis, with formation at each segment of analysis of discrete counts of current cepstrum of signal, through use of standard back Fourier transform ion segments one separates module values of discrete counts of current cepstrums of signal by segments of analysis. For each frequency of cepstrum of signal one forms variation series by segments of highlighted values of cepstrums, for each variation row of values of counts one highlights fractiles, with formation of relation of fractiles chosen on discrete values of frequency of cepstrum of signal, after that one compares dependences of fractiles obtained on discrete values of cepstrum of signal by Kolmogorovs criterion of coordination with respective dependences of fractiles on that very values of frequencies of cepstrums of standard realizationf of cepstrums of data base. The invention provides increase of effectiveness of systems of cepstral identification of language, hydro-acoustic, bio-medical, seismic, ultrasonic signals.