FIELD: computing technology.
SUBSTANCE: invention relates to the field of computing technology for processing audio data. The technical result is achieved by performing spectral analysis of the noise or the additive mixture of a speech signal and noise in a set interval — "sliding window", wherein said window is periodically shifted by an interval of a preset magnitude for the entire duration of analysis. Spectral analysis is performed by analysing the multi-frequency periodic signals represented by digital readouts using compensation of the sideband components. The harmonics are identified as attributed to a speech signal or interference based on the value of spectral density (SD), the number of simultaneous harmonics, the duration of existence thereof, and the dispersion of the values of SD harmonics. The information of the number of the speech signal, the values of frequencies, amplitudes, time of occurrence and duration of existence of the harmonics of the speech signal is supplied to the output of the system.
EFFECT: higher effectiveness of isolating harmonics of a speech signal in the presence of acoustic noise.
1 cl, 4 dwg
