基于小波包的皮肤听声器语音特征提取研究  

Voice feature extraction research of skin-hearing aid compact based on wavelet packet

作　　者：王钟斐[1] 王彪[1] 

机构地区：[1]宝鸡文理学院数学系,陕西宝鸡721013

出　　处：《电子设计工程》2011年第9期71-75,共5页Electronic Design Engineering

摘　　要：为了提高皮肤听声器分辨语音的能力,提出了利用小波包变换提取语言信号特征参数,并将其应用于皮肤听声器的新方法。小波包变换对频带进行多层次划分,提高了时-频分辨率,所以通过小波包变换得到的特征参数更加准确、更加精细,且是唯一的。最后,通过仿真实验说明了该特征参数的优越性,并且其在一定程度上提高了皮肤听声器分辨语音的能力,从而验证了该方法是可行的。To improve the ability of skin-hearing aid compact to distinguish the sound of speech, the new method using wavelet packet transform to extracte feature parameter from voice signal is proposed, which is applied to skin-hearing aid compact. By using wavelet packet transform to classify multi-level frequency band, time-frequency resolution is increased. Therefore, the obtained feature parameter by wavelet packet transform is more accurate and sophisticated, and if is unigue. Finally, experiments indicate the superiority of feature parameter. And it increases the ability of skin-hearing aid compact to distinguish the sound of speech, which verifies that the method is feasible.

关 键 词：小波包 皮肤听声 特征提取 唯一性 分辨 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]