帧同步混合小波包变换快速算法及其在语音处理中的应用

Frame-Synchronized Combined Wavelet Packet Transform Algorithms and Their Application to Speech Signal Processing

帧同步的小波包快速算法是一种类似 FFT的小波变换算法。本文提出帧同步混合小波包的分析方法 ,其特点是在发挥小波变换的时频分析特点的同时 ,分析时窗能够满足语音处理中帧长的要求 ,从而可以实现实时处理。这种混合小波包具有完全重构特性。文中用实验对比表明了帧同步混合小波包分析的优越特性。

The frame-synchronized wavelet packet transform algorithms can be used to analyze signals in a STFT-like frame-by-frame way. This paper uses wavelets with different filter lengths to generate a new wavelet packet called combined wavelet packet. It can reduce the analysis time-window length to fit the requirement of real-time speech analysis. Experimental results show that the wavelet packet transform can find out more speech details than STFT.