一种基于噪声分类的语音增强方法  

A Speech Enhancement Approach Based on Noise Classification

作　　者：袁文浩[1,2] 林家骏[1] 王雨[1] 陈宁[1] 

机构地区：[1]华东理工大学信息科学与工程学院,上海200237 [2]山东理工大学计算机科学与技术学院,山东淄博255049

出　　处：《华东理工大学学报（自然科学版）》2014年第2期196-201,共6页Journal of East China University of Science and Technology

基　　金：国家自然科学基金(61271349);华东理工大学基本科研业务费探索研究基金(WH1214015)

摘　　要：为了提高噪声估计的准确性,改进语音增强方法性能,在改进的最小控制递归平均算法(Improved Minima Controlled Recursive Averaging,IMCRA)的基础上提出了一种基于噪声分类的语音增强方法。该方法首先对含噪语音进行噪声类型的判断,然后根据判定的噪声类型选取相应的最优参数进行噪声估计,最后采用最优修正的对数谱幅度语音估计计算增强后的语音。该方法相对于传统IMCRA算法,在语音信号的还原和背景噪声的抑制两方面都有较好的性能。To improve the accuracy of noise estimate and the performance of speech enhancement algorithms,this paper proposes a noise-classification-based speech enhancement approach by using the improved minima controlled recursive averaging（IMCRA）algorithm.In this proposed algorithm,the type of noisy speech is firstly determined,which will be utilized to select the optimal parameters IMCRA to estimate the noise spectrum.Finally,the enhanced speech is calculated by the optimally modified logspectral amplitude speech estimator.Compared with the traditional IMCRA algorithm,the proposed approach can attain better performance in both the restoration of the speech signal and the suppression of background noise.

关 键 词：语音增强 噪声分类 噪声估计 

分 类 号：TN912.3[电子电信—通信与信息系统]