基于相关函数双麦克风语音增强算法的改进

发布时间：2018-05-30 19:59

本文选题：语音增强 + 双麦克风　；参考：《西安电子科技大学》2015年硕士论文

【摘要】：语音增强技术作为语音信号处理的一个重要组成部分在语音信号处理领域得到了广泛的应用。在实际的应用环境中噪声种类多种多样,而且噪声的存在损害了语音信号的可懂度和语音质量。单麦克风语音增强技术已经取得了许多研究成果,并且对一般环境下的加性噪声有着不错的处理效果。但是由于单麦克风在处理复杂噪声时性能比较差,因此双麦克风语音增强技术得到了迅速的发展并且逐渐成为一种主流的降噪算法。通过充分利用语音信号的空间信息,双麦克风语音增强算法实现了对语音信号的空间定位,以实现对目标语音的接收和噪声的滤除。常用的双麦克风语音增强算法有基于麦克风阵列的语音增强算法、基于相关函数的语音增强算法、基于相位差的语音增强算法以及基于功率差的语音增强算法。本文主要讨论了一种应用于助听器的基于相关函数的可以对语音信号中相干噪声进行有效滤除的双麦克风语音增强算法。并在此算法的基础上得到了一种应用于手持终端的基于相关函数的双麦克风语音增强改进算法。首先,为了使算法可以应用于手机中,本文在原有基于相关函数语音增强算法的基础上做了一系列改进。通过对原有的基于相关函数语音增强算法的分析,改进后的算法调整了对目标语音的接收方向,从而实现对与双麦垂直平面内语音信号的有效接收,并且滤除掉其它方向的干扰信号。通过结合一种最小跟踪算法和原始的谱减滤波算法,本文得到一种对非平稳噪声有更好滤除效果的新算法。并且通过一种基于估计信噪比的维纳滤波算法,进一步实现对静音区段内噪声的滤除。最后通过对固定平滑因子的分析发现其存在一些缺陷,因此本文通过一种经过优化的动态平滑因子来代替原来的固定平滑因子,这种动态平滑因子可以动态调整其估计值以便于更好的跟踪信号的真实值。本文通过模拟一种小型办公室仿真环境,对比了不同信噪比和不同噪声角度下单麦谱减法、原始双麦相关函数语音增强算法和改进双麦相关函数语音增强算法的降噪性能。通过对比发现改进后双麦相关函数语音增强算法性能相比较原始双麦相关函数语音增强算法得到了明显的提升,并且改进后的双麦语音增强算法实现了对特定角度语音信号的定向接收。因此本文经过改进的基于相关函数的双麦克风语音增强算法对手机应用环境有较好的适用性。
[Abstract]:As an important part of speech signal processing, speech enhancement technology has been widely used in the field of speech signal processing. In the practical application environment, there are many kinds of noise, and the existence of noise damages the intelligibility and speech quality of speech signal. Single microphone speech enhancement technology has made many achievements and has a good effect on additive noise in general environment. However, because of the poor performance of single microphone in dealing with complex noise, dual-microphone speech enhancement technology has been rapidly developed and gradually become a mainstream noise reduction algorithm. By making full use of the spatial information of the speech signal, the dual-microphone speech enhancement algorithm realizes the spatial localization of the speech signal, so as to realize the receiving of the target speech and the filtering of the noise. The common dual microphone speech enhancement algorithms include microphone array based speech enhancement algorithm, correlation function based speech enhancement algorithm, phase difference based speech enhancement algorithm and power difference based speech enhancement algorithm. This paper mainly discusses a dual microphone speech enhancement algorithm based on correlation function which can effectively filter the coherent noise from speech signal. On the basis of this algorithm, an improved dual-microphone speech enhancement algorithm based on correlation function is proposed for handheld terminals. First, in order to make the algorithm applicable to mobile phones, this paper makes a series of improvements on the basis of the original speech enhancement algorithm based on correlation function. Based on the analysis of the original speech enhancement algorithm based on correlation function, the improved algorithm adjusts the receiving direction of the target speech and realizes the effective reception of the speech signal in the vertical plane with the double wheat. And filter out interference signals in other directions. By combining a minimum tracking algorithm with the original spectral subtraction filtering algorithm, a new algorithm with better filtering effect for non-stationary noise is obtained in this paper. Furthermore, a Wiener filter algorithm based on the estimation of SNR is used to remove the noise in the mute region. Finally, it is found that there are some defects in the fixed smoothing factor, so this paper uses an optimized dynamic smoothing factor to replace the original fixed smoothing factor. This dynamic smoothing factor can dynamically adjust its estimation to better track the true value of the signal. In this paper, by simulating a small office simulation environment, we compare the noise reduction performance of different signal-to-noise ratio (SNR) and different noise angle order wheat spectrum subtraction, original double wheat correlation function speech enhancement algorithm and improved double wheat correlation function speech enhancement algorithm. It is found that the performance of the improved algorithm is better than that of the original one. And the improved algorithm realizes the directional reception of the specific angle speech signal. Therefore, the improved dual microphone speech enhancement algorithm based on correlation function is suitable for mobile phone application environment.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN912.35

【相似文献】