Description: The relevance of a scientific task, audio signal extraction from the background of noise, with using the method of the singular spectral analysis (SSA) will be proved in this paper. The SSA method is based on the transformation of the original onedimensional time series into a multidimensional vector representation, followed by processing by the algorithm of the additive principal components method, which makes it possible to study stationary and non-stationary time series without a priori information. Application of this method allows to extraction low-frequency main components containing information in the audio signal, which is distorted by additive noise. The algorithm of the mathematical model of audio signal extraction from the background of noise using the singular spectral analysis method is proposed. Quantitative graphs obtained as a result of modeling a mathematical model of the method of singular spectral analysis, which characterize the properties of the extraction of an audio signal distorted by noise with different intensities, are studied. The obtained results testify to the possibility of an audio signal extraction with a ratio of the root-mean-square deviations of the signal-to-noise ratio of not less than 0.5, and also with a correlation coefficient between the original and the selected signals of not less than 0.82. This method is applicable to increase the noise immunity of avionics and the quality of speech intelligibility in the investigation of aviation accidents.
Keywords: singular spectral analysis, audiosignal extraction, noise, noise immunity, avionics