基于动态谱估计的改进谱减语音增强算法  

Improved spectrum subtraction speech enhancement algorithm based on dynamic spectrum estimation

作　　者：陈武[1] 朱忠[1] 陈琳[1] 李强[1] 

机构地区：[1]中国卫星海上测控部,江苏江阴214431

出　　处：《电子设计工程》2014年第1期35-37,共3页Electronic Design Engineering

摘　　要：语音增强是语音信号处理的重要课题。根据基于最小值追踪的谱估计方法,提出了一种非平稳噪声环境下快速追踪噪声变化的方法,将其应用到改进后的谱减法中,以提升语音增强的效果。仿真结果表明,改进后的谱减法能有效降低背景噪声,提高输出语音信号的信噪比。Speech enhancement is important topic in speech signal processing. According to the spectrum esumauon method based on minimum track, proposed a fast track method for non-stationary noise environment, and apply it to the improved spectrum subtraction, to enhance the effect of speech enhancement. Simulation results show that the improved spectrum subtraction is effective in reducing background noise, improving the Signal Noise Ratio（SNR） of the output speech signal.

关 键 词：语音增强 谱估计 噪声 谱减法 

分 类 号：TN912[电子电信—通信与信息系统]