Psychoacoustic active noise control system based on empirical mode decomposition

A psychoacoustic active noise control (ANC) system based on empirical mode decomposition (EMD) is proposed and implemented to improve the noise reduction performance of the control system. The noise source signal is decomposed by EMD, and the psychoacoustic parameter â–œloudnessâ–? of each intrinsic mode function (IMF) is initially calculated in such a system. Thereafter, the high-pass psychoacoustic weighting filter used to shape the error and reference signals is designed adaptively and automatically according to the loudness, peak frequency, and amplitude of each IMF. Three different ANC systems are simulated, and the sound pressure levels and loudness of their residual error signals are compared. The results demonstrate that the filter designed using this method can restrain the components of noise sources with small loudness better than the A-weighting shaping filter, so that the proposed control system can improve the noise reduction compared to those of the filtered-x least mean square and A-weighting shaping filters. Finally, the computational complexity of the three ANC systems is analyzed and compared.