Аналіз звукового спектру

Abstract

This paper focuses on the mathematical foundations and practical implementation of audio spectrum analysis methods in digital signal processing systems. The study provides a comprehensive review of the Fast Fourier Transform algorithm and its application for decomposing non-stationary audio signals into frequency components. Special attention is paid to the problem of spectral leakage and the effectiveness of various window functions, such as Hann, Hamming, and Blackman windows, in mitigating this artifact. The research also explores the construction of spectrograms for time-frequency analysis and the extraction of Mel-frequency cepstral coefficients for speech recognition tasks. Experimental results confirm that the proposed adaptive windowing approach significantly improves frequency resolution and reduces side-lobe levels in real-time audio processing applications