噪声环境下语音信号增强方法和研究  

Method and Research of Speech Signal Enbancement in Noisy Environment

作　　者：李思嘉[1] 赵婧 蔡树阳 LI Sijia;ZHAO Jing;CAI Shuyang(College of Software,Henan Normal University,Xinxiang 453007,China;College of Computer Information Engineering,Henan Normal University,Xinxiang 453007,China)

机构地区：[1]河南师范大学软件学院,河南新乡453007 [2]河南师范大学计算机与信息工程学院,河南新乡453007

出　　处：《计算机与网络》2022年第8期60-63,共4页Computer & Network

基　　金：国家自然科学基金(61040037)。

摘　　要：基于小波去噪理论实现了阈值法语音增强算法,传统的软阈值法与硬阈值法在去噪处理上均有不足,经过软阈值法处理的语音信号与原始信号相比存在恒定误差,经过硬阈值法处理的语音信号不够光滑,且附加震荡的存在。在软硬阈值法的基础上提出了一种改进的软硬阈值折衷算法,提出的改进软硬阈值折衷算法和传统的阈值算法相比其适应性更优,克服了硬阈值函数不连续的缺点,减小了软阈值函数中估计小波系数与分解小波系数的恒定偏差,仿真结果说明,根据噪声信号的不同,通过修改式中系数,可得到比传统阈值法去噪更优的结果。Based on the wavelet denoising theory,the threshold speech enhancement algorithm is realized.The traditional soft threshold method and hard threshold method have shortcomings in denoising.The speech signal processed by soft threshold method has constant errors compared with the original signal,while the speech signal processed by hard threshold method is not smooth enough,and there is additional oscillation.Based on the soft and hard threshold method,an improved soft and hard threshold compromise algorithmis proposed.Compared with the traditional threshold algorithm,the proposed soft and hard threshold compromise algorithm has better adaptability,overcoming the discontinuity of the hard threshold function and reducing the constant deviation between the estimated wavelet coefficients and the decomposed wavelet coefficients in the soft threshold function.The simulation results showthat according to the different noise signals,the denoising results can be better than the traditional threshold method by modifying the coefficients in the formula.

关 键 词：语音去噪 语音增强 阈值函数 小波变化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]