基于概率耦合的双直接判决先验信噪比估计算法  

Probabilistic Combination Framework of Two Decision-Directed Algorithms for a Priori SNR Estimation

作　　者：欧世峰[1] 赵艳磊 宋鹏 高颖[1] OU Shi-feng;ZHAO Yan-lei;SONG Peng;GAO Ying(School of Science and Technology for Opto-electronic Information,Yantai University,Yantai,Shandong 264005,China;School of Computer and Control Engineering,Yantai University,Yantai,Shandong 264005,China)

机构地区：[1]烟台大学光电信息科学技术学院,山东烟台264005 [2]烟台大学计算机与控制工程学院,山东烟台264005

出　　处：《电子学报》2020年第8期1605-1614,共10页Acta Electronica Sinica

基　　金：山东省自然科学基金(No.ZR2017MF008)。

摘　　要：直接判决(DD,Decision-Directed)算法结构简单、音乐噪声抑制能力较好,是当前语音增强领域最为常用的先验信噪比估计方法.但该算法对于滑动因子的选取数值较为敏感,且估计性能要受到时延问题的限定.本文首先采用实际的语音和噪声数据,根据音乐噪声残留及输出语音失真两方面的评测标准对DD算法中滑动因子的取值问题进行了研究,通过数据分析给出了其较为明确的上下边界值;然后基于语音及噪声信号的复高斯分布模型,采用软判决技术对两个具有不同滑动因子的DD算法进行概率耦合,提出了一种具有双DD结构的先验信噪比估计算法.该算法可以充分结合两个具有不同特性DD算法的优点,在音乐噪声抑制及限制语音失真等方面均获得了较为理想的输出效果.多种噪声背景及输入信噪比条件下的仿真结果表明,相对于目前流行的几种先验信噪比估计算法,本文提出算法具有更为优良的估计性能.Due to the low computational complexity and acceptable ability in reducing musical noise effect,the decision-directed(DD)approach is widely used for estimating the a priori signal-noise-ratio(SNR)in many speech enhancement systems.However,the DD approach suffers from the problem of time delay and the performance is very sensitive to the fixed smoothing factor.Firstly,the performance of DD approach in musical noise reduction as well as speech distortion attenuation are analyzed using actual speech and noise data,and the boundary values of smoothing factors are presented in view of the analyzed results.Then,a novel algorithm is proposed,in which two DD approaches with different smoothing factors are probabilistically combined in an attempt to put together the best properties of them.The contribution of either DD approach to the combination is automatically adjusted in accordance with the speech absence probability,which can be computed using the complex Gaussian model and soft decision technique.Experiments are carried out in different noise and input SNR conditions,and the results demonstrate that the proposed algorithm can significantly outperform the popular methods for estimating the a priori SNR.

关 键 词：语音增强 直接判决算法 滑动因子 概率耦合 

分 类 号：TN912.3[电子电信—通信与信息系统]