基于盲信号分离的自适应回声抵消算法

在视频会议和免提通信系统中,扬声器和麦克风之间的声耦合严重影响语音通信系统的质量。文中提出了一种用基于盲信号分离（BSS）的自适应声回波对消（AEC）方法,可有效解决回声和噪声对近端语音信号的影响。该方法不仅能减少背景噪声对回波对消的影响,而在双向通话时,可直接利用盲信号分离技术分离出近端语音。     

一个基于盲信号分离的多路回波消除结构

多路回波消除问题的本质难点是由于多路输入信号之间的强相关性而导致的自适应滤波器解不唯一的问题.目前许多文献都是试图通过对输入信号预处理来解决这个问题,这些方法都对原输入信号作了变化,从而影响了通话质量.本文提出了一个简化的盲信号分离模型的立体声回波消除结构,在这个结构中,立体声回波消除问题被转化为两个单路的回波消除问题,从而避免了自适应滤波器解不唯一的问题.仿真结果证明了新结构的有效性.     

一种有效的语音盲信号分离简化混合模型

本文针对语音盲信号分离问题,提出了一种新的简化混合模型.基于从同一声源到距离靠近(距离大约在几十厘米的范围内)的两个麦克风的房间冲激响应具有相似性这一特点,该模型只需利用这两个冲激响应的差异信息,因而简化了盲源分离的计算复杂度.自行设计实现的语音盲信号分离实时DSP实验系统证实,该模型有效可行,特别是通过引入时间延迟来解决非最小相位传递函数的求逆问题使得盲信号分离性能得到显著的改善.     

基于盲信号分离算法的立体声回波抵消的研究

当处理混合语音信号时,传统的立体声回波抵消技术不具备修正参数的能力.因此,传统的立体声回波抵消技术对于信号传输的可靠性存在很大的问题,论文中,我们提出一种基于盲信号分离方法(BSS)的立体声回波抵消技术.在我们提出的系统中,我们通过分离回波信号和说话人的语音信号来达到移除回波信号的目的.结果证明,本方法对于传输信号的可靠性有明显的提高.     

超定盲信号分离RLS算法研究

研究在线盲信号分离问题.将适定盲信号分离的RLS算法,推广到源信号个数未知的超定盲信号分离模型中,利用正交投影方法消除了超定盲信号分离RLS算法的冗余移动,设计出能够稳定收敛的超定盲信号分离RLS算法.仿真实验验证了算法的有效性和其收敛的稳定性.     

粒子群算法用于盲信号分离的研究

提出一种采用粒子群优化算法进行盲信号分离的新方法,为盲信号分离领域提供一种新的研究思路与方法。该方法采用峰度作为适应度函数,利用粒子群算法对由多个源信号混合而成的信号进行盲信号分离。与自然梯度法盲信号分离相比,粒子群算法精度更高,收敛速度更快,实例仿真成功地对两个图像混合信号进行了盲分离,表明了算法的有效性和优越性。     

基于盲信号分离的同频信号的串行分离技术

传统的时域和频域处理等方法很难分离多个频域的混叠信号,文中提出一种利用盲信号分离技术串行分离同频信号的新方法——FastICA盲分离算法。该方法充分利用盲信号分离技术不需要知道信号先验信息的特性,能正确地、一个一个地分离出在频域中混合在一起的信号,且能分离功率相差100万倍的同频信号。在存在比信号功率大得多的高斯白噪声情况下,FastICA盲分离算法具有良好的分离性能。     

空间电子探测信号盲分离研究

给出了盲信号分离中的瞬时混合,时延混合和卷积混合三种混合模型,介绍了两种具体的盲分离算法,等变自适应盲分离算法和非高斯性最大化的快速定点算法.其中对于窄带源信号,对时延混合模型进行了扩展,提出了用复数域瞬时盲信号分离算法分离时延混合信号的新思路.最后给出了相应的仿真和实验结论,实验结果表明用基于复数的盲分离算法确实能够有效地分离阵列接收的时延混合信号.     

一种基于ICA的盲信号分离快速算法

基于ICA(独立成分分析:Independent Component Analylsis)原则,给出一种盲信号分离的快速学习算法.通过寻求观测变量线性组合的四阶累积量(即kurtosis系数)局部极值,得出该算法的模型和步骤.将该算法用于盲信号分离实验,实验结果表明,该算法在盲信号分离和信号特征提取方面具有收敛速度快、无需动态参数等优点.该算法能有效地分离出任意分布的非高斯盲源信号的各个独立成分,是信号处理的一种新的、高效可靠的方法.     

盲信号分离及盲信号抽取研究

盲信号分离技术是将混合信号中的源信号分离出来的一种功能强大的信号处理方法,已成为信号处理领域的研究热点。阐述了盲信号分离的发展现状,介绍了盲信号分离问题的数学模型,给出了盲源分离的基本思想。对盲信号分离算法进行了研究,阐述了盲信号分离几种典型算法的特点及性能,对与盲信号分离紧密相关的盲信号抽取算法进行了总结,并对盲信号分离的进一步研究进行了展望。     

Adaptive Sampling Rate Correction for Acoustic Echo Control in Voice-Over-IP

Hands-free terminals for speech communication employ adaptive filters to reduce echoes resulting from the acoustic coupling between loudspeaker and microphone. When using a personal computer with commercial audio hardware for teleconferencing, a sampling frequency offset between the loudspeaker output D/A converter and the microphone input A/D converter often occurs. In this case, state-of-the-art echo cancellation algorithms fail to track the correct room impulse response. In this paper, we present a novel least mean square (LMS-type) adaptive algorithm to estimate the frequency offset and resynchronize the signals using arbitrary sampling rate conversion. In conjunction with a normalized LMS-type adaptive filter for room impulse response tracking, the proposed system widely removes the deteriorating effects of a frequency offset up to several Hz and restores the functionality of echo cancellation.     

Combined acoustic echo cancellation,dereverberation and noise reduction: a two microphone approach

This paper presents new algorithms for acoustic echo cancellation and noise reduction which use two (or possibly more) microphone signals. In contrast to the single microphone method the multimicrophone approach can exploit the spatial coherence properties of sound fields which arise from noise and reverberated speech. Besides the standardfir echo canceller the proposed algorithms comprise an adaptive filter to eliminate non coherent signal components. The combined system achieves better Erle than thefir echo canceller alone, attenuates ambient noise, dereverberates near end speech, and possibly leads to implementations with reduced complexity. The paper analyzes the acoustical properties of typical environments, presents the algorithms and experimental results.     

Exploiting Spin Echo Decay in the Detection of Nuclear Quadrupole Resonance Signals

Nuclear quadrupole resonance (NQR) is a radio-frequency technique that can be used to detect the presence of quadrupolar nuclei, such as the ¹⁴N nucleus prevalent in many explosives and narcotics. In a typical application, one observes trains of decaying NQR echoes, in which the decay is governed by the spin echo decay time(s) of the resonant line(s). In most detection algorithms, these echoes are simply summed to produce a single echo with a higher signal-to-noise ratio, ignoring the decaying echo structure of the signal. In this paper, after reviewing current NQR signal models, we propose a novel NQR data model of the full echo train and detail why and how these echo trains are produced. Furthermore, we refine two recently proposed approximative maximum-likelihood detectors that enable the algorithms to optimally exploit the proposed echo train model. Extensive numerical evaluations based on both simulated and measured NQR data indicate that the proposed detectors offer a significant improvement as compared to current state-of-the-art detectors     

Geometric acoustic attenuation at gigacycle frequencies in single crystals of quartz

Successive acoustic echo amplitudes have been calculated from a model for microwave frequency acoustic propagation based on ray optics that ignores any intrinsic attenuation. The computed echoes resemble experimental acoustic decay patterns generated by microwave pulses in single crystals of piezoelectric materials. The attenuation has been found to be due to off-axis propagation of the acoustic energy in the crystal and to acoustic phase averaging by the microwave transducers used to couple energy in and out of the crystal. Calculated acoustic echo patterns show the changes caused by various crystal orientations and different crystal endface parallelisms at several frequencies. It is found that the number of echoes and the amplitude of the last few echoes are frequency independent and are determined by the crystal geometry alone. On the other hand, the character of all but the last few echoes is dominated by phase averaging effects over the crystal end wall and is strongly frequency-dependent. Experimental data is presented showing the changes in the nulls and maximums of the echo patterns with frequency and the changes in acoustic patterns produced by altering the endface parallelism by flexing the crystal. Measured acoustic echo patterns of flexed and unflexed crystals are discussed. It is shown that flexing has increased the number of observable echoes by factors from 4 to 10 for various samples, has improved the average intensity of comparable echoes by amounts varying from 4 to 14 db, and has reduced the apparent acoustic attenuation from an unflexed value of 1.86×10^-1db/cm to 1.63×10^-2db/cm for a particular sample.     

Multipath delay estimation using a superresolution PN-correlationmethod

This paper addresses the problem of high-resolution estimation of a multipath channel delay profile. We propose several improvements to the so-called superresolution pseudo-noise sequence correlation method (SPM) and analyze its performance on time-varying channels. SPM is based on the multiple signal classification (MUSIC) algorithm, which requires decorrelation of the multipath echoes. The proposed improvements enable SPM-based delay estimation in the presence of narrowband interference, and they reduce the necessary transmission window while preserving multipath echo decorrelation. These improvements are analyzed and are applied to underwater acoustic experimental data     

基于盲源分离理论的麦克风阵列信号有音/无音检测方法

该文提出一种在方向性噪声场中多路麦克风信号同时进行有音/无音检测(VAD)的方法。在方向性噪声场中,由于各个麦克风接收信号中的噪声彼此之间相关,因而,可以利用盲源分离理论将方向噪声与语音源信号分离,从而获得相对比较纯净的语音源信号。对分离出的语音源信号进行有音/无音检测,获得VAD结果,同时估计出各个麦克风信号相对于该信号的时延值。以相对纯净语音源信号的VAD检测结果为参考,将其分别平移相应的时延值,即可同时获得多路麦克风信号的VAD结果。计算机模拟结果表明,在方向性噪声场的多种情况下,该方法对具有加性噪声的多路麦克风信号均具有较好的有音/无音检测能力。     

基于梯度向量投影的自适应滤波算法的改进及其在多路回波消除中的应用

该文首先对Lim(2000)的基于梯度向量正交投影的算法(OGA)进行了分析和改进,在此基础上获得了一种新的自适应滤波算法(MOGA)。新算法使用时变遗忘因子对误差进行指数加权平均来估计均方误差,并使用该因子改变自适应迭代过程中滤波器系数向量的更新方向.然后将改进后的新算法扩展成两路回波消除算法用于多路回波的消除中,获得了良好的效果。仿真结果表明, MOGA不仅对时变或时不变系统具有很好的跟踪能力,克服了Lim(2000)所提算法收敛性不佳甚至有时发散的缺陷,而且应用于多路回波消除时具有计算量小,收敛速度快和精度高等特点,其收敛速度和精度优于J.Benesty(1996)和G.Sankaran(1999)的相应结果。     

基于LTE信号的小目标回波相参积累:目标散射建模与影响分析

长时间的信号积累是提高微弱慢速小目标探测的一种有效方法,但积累时间长,目标回波会发生越距离单元和多普勒单元走动,相参积累困难。本文对基于LTE信号的小目标探测问题开展研究,引入目标散射空变性,建立了基于LTE通信体制的系统回波模型;在此基础上,分析了目标回波距离走动和多普勒走动特性,提出了基于keystone变换和分数阶傅里叶变换（FRFT）的回波相参积累技术;基于典型目标散射模型,分析了目标散射特性对积累效果的影响,并在典型系统条件下开展了仿真实验,验证了所提算法和分析的有效性。