一种基于短时谱估计和人耳掩蔽效应的语音增强算法  

Speech Enhancement Based on Masking Properties and Short-Time Spectral Amplitude Estimation

作　　者：陈国明[1] 赵力[1] 邹采荣[1] 

机构地区：[1]东南大学无线电工程系,南京210096

出　　处：《电子与信息学报》2007年第4期863-866,共4页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(60472058);教育部博士基金(20050286001);教育部新世纪优秀人才支持计划(NCNE-04-0483)资助课题

摘　　要：该文结合短时谱估计算法和人耳掩蔽效应提出了一种单通道语音增强算法。该算法在MMSE准则下采用了非固定参数的语音跟踪,并且引入人耳掩蔽效应动态的确定增强滤波器的传递函数以适应语音信号的变化。实验结果表明:该算法使降噪后的语音信号有较小的语音失真并且很好地抑制了音乐噪声。In this paper a single channel speech enhancement method for noisy speech signals is proposed, which is based on masking properties of the human auditory system and power spectral density estimation. During the estimation of power spectrum of the speech, the estimator can be modified by the MMSE method and the estimator of short -time spectral amplitude filter is determined by masking properties. In this way, the best trade off among the reduction of noise, the speech distortion and the level of musical residual noise can be found. Experimental results demonstrate the improved performance of the proposed algorithms.

关 键 词：语音增强 听觉掩蔽 音乐噪声 

分 类 号：TN912.3[电子电信—通信与信息系统]