In the thesis, some improvements are made to the time-frequency transform and the psychoacoustics model of MPEG. As to the block of time-frequency, division of wavelet packet to signal frequency bands is consistent with the division of critical bands as possible by adopting the wavelet packet transform, which can get better frequency resolution at low frequency and better time resolution at high frequency, and get higher compression ratio. As to the psychoacoustics model, psychoacoustics model is established in the wavelet domain, the masking threshold is computed directly in the wavelet domain, which need not time-frequency analyzing, which improve the algorithms and computing complexity.This improved algorithm is programmed using C language and simulated in Matlab environment. Compared with the standard algorithm, the improved algorithm makes some improvement to perceptual effect, compression ratio and computing complexity.
|