Perceptual Filters for Audio Signal Enhancement
D. Tsoukalas, John Mourjopoulos, George C. Kokkinakis · Journal of the Audio Engineering Society · 1997
A novel technique for the removal of broad-band noise from audio signals is presented. The technique relies on a psychoacoustic representation of the signals, which allows the introduction of a linear time-frequency-varying perceptual enhancement filter. It is shown that the performance of this filter can be optimized using the frequency- and time-domain properties of the psychoacoustic model. It is also shown that in terms of signal enhancement, the proposed perceptual filter is a very powerful tool, capable of both removing and psycho-acoustically shaping all perceptually significant noise. The proposed filter is evaluated with a number of objective criteria, which assess its noise suppression capabilities together with possible degradations imposed on the processed signal.