Fast and Robust DOA Estimation Method Using Searching Space Pre-estimation

An improved MUSIC localization algorithm is proposed in this paper in order to locate a sound source in reverberant and directional noise scenario, with comparatively low computational complexity. Based on Time delay of arrival （TDOA） estimation, a candidate searching space is first predetermined and then Beamspace MUSIC combined with Spatial smoothing （BMUSIC-SS） algorithm is applied in this space to find the Direction of arrival （DOA） of the desired source. This two-step method not only reduces the number of calculations by avoiding the exhaustive search, but also improves the localization performance in coherent noise environment due to the fact that the spaces including directional noises are partially removed. Besides, based on the localization system we developed, the DOA parameters are separated to reduce a two-dimension search into a couple of one dimensional ones that further cuts down the computational load. The computer simulations and practical experiments demonstrate the robustness and efficiency of the proposed method.     

麦克风阵列声源定位系统的研究

随着多媒体技术的进一步发展,语音接收和声音信号处理得到了日益广泛的关注和应用,而声源的定位和声源增强是实现语音增强,语音识别的前提和基础.基于麦克风阵列的声源定位技术由于其广泛的应用前景得到了众多学者的关注.单个麦克风接收到的信息量较少,缺少声源定位所需要的信息,而麦克风阵列克服了上述的缺点,利用了各麦克风信号之间信号的相关性对数据进行相关分析和处理从而实现声源的定位.文中阐述了麦克风阵列声源定位的原理,推导计算目标方位角、俯仰角以及距离的计算公式;阐述了硬件系统的组成以及各个部分的作用并通过实验进行了系统的测试,通过对测试数据的分析得出麦克风阵列声源定位系统能够实现声源的快速定位.     

基于盲波束形成的麦克风阵列语音增强方法

考虑到语音信号方向向量估计误差对传统波束形成语音增强性能的影响,该文提出一种盲波束形成语音增强方法。由于采用阵列四阶互累积量和线性约束最小方差波束形成器相结合,使得该方法对语音信号方向向量误差具有一定韧性。此外,采用多通道后置滤波去除盲波束形成器输出端的残留噪声。仿真结果表明,在语音信号波达方向等先验信息未知的情况下,该文提出的盲波束形成语音增强方法仍具有较好的噪声抑制性能。     

Acoustic emission monitoring during laser shock cleaning of silicon wafers

A laser shock cleaning is a new dry cleaning methodology for the effective removal of submicron sized particles from solid surfaces. This technique uses a plasma shock wave produced by laser-induced air breakdown, which has applied to remove nano-scale silica particles from silicon wafer surfaces in this work. In order to characterize the laser shock cleaning process, acoustic waves generated during the shock process are measured in real time by a wide-band microphone and analyzed in the change of process parameters such as laser power density and gas species. It was found that the acoustic intensity is closely correlated with the shock wave intensity. From acoustic analysis, it is seen that acoustic intensity became stronger as incident laser power density increased. In addition, Ar gas has been found to be more effective to enhance the acoustic intensity, which allows higher cleaning performance compared with air or N₂ gas.     

改进的后滤波波束形成器语音增强算法

该文提出了一种具有后滤波的波束形成器的语音增强改进算法。该算法主要解决维纳滤波器的理想信号功率谱估计,结合自功率谱减法和互功率谱减法计算出尽可能多的功率谱估计值,以使平均结果更接近于真实值,同时修正了声源移动引起的互功率谱变化。实验结果信噪比提高5dB以上,汽车环境中基于隐含马尔可夫模型(HMM)的小词汇量短语识别达到84％。从信噪比、平均谱距离和语音识别率可以看出该算法有效去除了原始算法中易残留的低频噪声,减少了语音信号失真。     

2001年初长沙市场家庭音响产品质量监督检测结果分析

通过对长沙市市场随机抽取的四种产品：视盘机、功放、音箱、话筒100个样品进行质量监督，对不合格产品进行了分析。     

浅析通俗音乐中人声的录制和处理方法

介绍了录制某类人声所使用的一般传声器和摆放原则,以及后期处理的一般方法和手段.通过获得贵州省新闻出版广电局2014年广播节目录制技术质量奖一等奖作品——《爱在路上》为例,分析和总结了通俗音乐人声录制与处理方法的特点与规律.     

超声速弹丸立靶弹着点定位模型研究

针对常规武器火炮直瞄射击时的弹着点测试问题,提出了一种双圆环形传声器阵列方法。该阵列将若干传感器布设在两个圆环上,并根据超声速弹丸产生的激波传播速度随传播距离增加逐渐递减的规律,对每个传声器阵分别引入一个视速度,建立了双视速度模型。计算机仿真分析表明,采用双圆环形传声器阵列及双视速度模型确定弹丸弹着点坐标,带来的模型误差优于3 mm,对于工程应用而言可以忽略。试验结果表明,双视速度模型明显优于将激波传播速度视为恒定的基本模型和交会模型。     

新型集成微麦克风和扬声器的结构设计

在现有的硅基Ｐｂ（Ｚｒ,Ｔｉ）Ｏ３（ＰＺＴ）铁电薄膜的制备工艺基础上,提出了一种以ＰＺＴ悬臂式振膜为核心的集成微麦克风和扬声器结构,对其进行了优化设计。对微麦克风和扬声器的灵敏度和声输出进行了理论计算。并比较了基于ＰＺＴ和ＺｎＯ两种材料的微麦克风和扬声器的性能差异。