基于麦克风阵列的多声源测向方法研究

发布时间：2018-04-10 00:35

本文选题：麦克风阵列　切入点：多源测向　出处：《南京理工大学》2014年硕士论文

【摘要】：基于麦克风阵列的多声源测向技术通过对麦克风阵列接收的多声源混合信号进行处理,从而确定各个声源的方位。它在很多领域都具有广泛的应用前景和实际意义,如在民用方面的视/音频会议、语音识别及增强等领域中,常利用声源测向技术精确估计出说话人位置来控制摄像头,使其自动对该位置的语音信号进行增强。在军事方面声源测向技术被广泛地应用在飞机,火炮、狙击手探测等方面。因此,该技术成为了语音信号处理领域的研究热点之一。本课题针对基于麦克风阵列多声源测向问题展开研究,归纳总结并比较了传统的几类声源测向方法。本文以典型的双阵元麦克风阵列为研究对象,针对远场多声源模型,将基于语音信号时频正交特性的退化分离估计技术(DUET)应用于声源信号测向。该算法利用了语音信号特有的时频稀疏和短时正交特性(W-Disjoint Orthogonality, W-DO),基于此特性的时延估计算法计算量小,实现简单,仅用两个麦克风就可以实现多个声源的方位测向。但是当声源存在波长小于两倍阵元间距的高频成分时,此类声源测向方法将出现相位卷绕模糊问题,而阵元间距因物理尺寸限制也不可能无限缩小,因此限制了该类方法的实际应用领域。针对上述问题,本文提出了一种基于迭代时频掩蔽的宽间距麦克风阵列多声源测向方法,该方法通过迭代消去过程,显著抑制了相位卷绕产生的影响。此外,结合基于能量的语音端点检测技术,本文进一步给出了上述方法的实时处理算法步骤。针对上述方法,本文进行了仿真实验和相关外场实验,实验结果表明：针对宽间距麦克风阵列多声源测向,本文所述方法明显优于常规DUET类方法,具有一定的实际应用价值。
[Abstract]:Multi-sound source direction-finding technology based on microphone array processes the mixed signals received by microphone array to determine the orientation of each sound source.It has a wide application prospect and practical significance in many fields, such as video / audio conference, speech recognition and enhancement in the civilian field, it often uses the sound source direction finding technology to accurately estimate the speaker's position to control the camera.Make it automatically enhance the speech signal at this position.Sound source direction finding technology is widely used in aircraft, artillery, sniper detection and so on.Therefore, this technology has become one of the research hotspots in the field of speech signal processing.This paper focuses on the research of multi-source direction finding based on microphone array, and summarizes and compares several traditional sound source direction finding methods.In this paper, a typical dual-array microphone array is studied. For the far-field multi-source model, the degenerate separation estimation technique based on the time-frequency orthogonality of speech signal is applied to the direction finding of the sound source signal.This algorithm utilizes the time-frequency sparse and short-time orthogonal characteristic of speech signal, W-Disjoint Orthogonality (W-DOA). The time delay estimation algorithm based on this characteristic has the advantages of small computation and simple implementation, and it can realize the azimuth direction finding of multiple sound sources with only two microphones.However, when the sound source has a high frequency component whose wavelength is less than two times the spacing of the array element, the phase winding ambiguity will occur in this method, and the distance between the elements can not be reduced indefinitely because of the limitation of the physical size.Therefore, the practical application field of this kind of method is limited.In order to solve the above problems, an iterative time-frequency masking method for direction finding of multi-source microphone arrays with wide spacing is proposed. The effect of phase winding is significantly suppressed by iterative elimination process.In addition, combined with the energy-based speech endpoint detection technology, this paper further gives the real-time processing algorithm steps of the above method.In view of the above methods, the simulation experiments and related field experiments are carried out in this paper. The experimental results show that the method presented in this paper is obviously superior to the conventional DUET method and has certain practical application value for the wide spacing microphone array multi-sound source direction finding.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.3

【参考文献】