Оптимізація спектрального гейтінгу для адаптивного шумозаглушення

Abstract

Digital audio processing is essential for professional recording, bioacoustics, and speech recognition, where background noise often degrades signal quality. Spectral gating helps reduce noise by separating harmonic components from unwanted acoustic artifacts using the Fast Fourier Transform (FFT), which converts audio from the time domain to the frequency domain. By optimizing FFT window size and using parameters such as the over-subtraction factor (α) and spectral floor (β), systems can balance noise reduction with the preservation of natural sound