基于扩展谱减法的语音增强算法及实现  

Algorithm and Realization of Speech Enhancement Based on Extended Spectral Substraction

作　　者：李媛[1] 铁勇[1] 那顺乌力吉[1] 李树华[1] 王强[1] 

机构地区：[1]内蒙古大学理工学院电子工程系,呼和浩特010021

出　　处：《内蒙古大学学报（自然科学版）》2008年第1期97-101,共5页Journal of Inner Mongolia University：Natural Science Edition

基　　金：国家自然科学基金(60772166);内蒙古自然科学基金(200711020811)

摘　　要：结合漏水信号的特性,对基本谱减法进行改进.加入重叠帧,抑制部分音乐噪声.为了适合漏水信号无静寂帧的特点,引进了由谱减法和维纳滤波器相结合而构成的扩展谱减法,使得噪声估计可以时刻更新.经仿真验证,滤噪效果更好,适合用于对漏水信号的滤噪处理.In order to improve basic spectral substraction,based on the characteristic of ieak water signal, the overlap frame is joined to reduce the influence of music noise, and the spectral substraction and wiener filter consisting of extended spectral substraction are used to suit the leak water signal which does not have silence frame. This causes the estimate of noise renewing frequently. Confirming after the emulation,the noise filtered with extended spectral substraction is better. It is suitable to the leak water signal of noise filtered processing.

关 键 词：音乐噪声 静寂帧 扩展谱减法 

分 类 号：TN912.35[电子电信—通信与信息系统]