小波域心理听觉模型

Psycho-acoustic model in the wavelet domain

提出一个简单直接的将频域心理听觉模型转换到小波域,进行小波包滤波器分解音频编码的新方法.该方法利用离散傅里叶变换左右频带的对称性,将频域的掩蔽域值频谱作为一个信号,利用傅里叶反变换和离散小波包变换转换到每一个小波子带,根据掩蔽信号在小波域的能量来控制子带中音频信号的量化步长,利用该方法实现小波包音频编码.

A simple and direct algorithm to convert the psycho-acoustic model from the frequency domain to wavelet domain is presented to perform the Discrete Wavelet Packet Transform. This algorithm uses the symmetry of right half-band frequency spectrum and left half-band frequency spectrum in the process of DFT（ Dis- crete Fourier Transform）, and makes the masking threshold in the frequency domain as a signal. The inverse DFT and DWPT（ Discrete Wavelet Packet Transform） are used to convert the masking signal into the wavelet domain. The wavelet coefficients＇ quantization stepsize of audio signal is determined according to the masking signal energy in the wavelet packet Subband, and then wavelet packet audio coding is realized.