【作者】 商文忠；

【导师】 姬红兵；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2007， 硕士

【摘要】 方向性技术是目前解决助听器用户在噪声环境下语言理解困难最有效的方法之一,主要包括声源定位与噪声源跟踪。针对目前国内己有方向性技术存在的精度不高、运算量较大、难以实时处理等问题,本文深入系统地研究了基于麦克风阵列的声源定位和噪声源跟踪方法,给出了一种自适应方向性系统。首先,论文研究了基于时延估计的声源定位方法。并对两种经典算法进行了改进:对于频域最小均方自适应滤波时延估计,在相位数据拟合时,进行一次加权处理,有效地改善了时延估计精度和自适应收敛速度;对于自适应特征值分解时延估计,结合LMS时延估计的新方法能够更快地收敛到新的时延。其次,研究了一种有效的自适应方向性系统。该系统基于两个固定极性的自适应组合,使系统输出的极性零点总是朝向噪声方向,并给出了相关更新增益的自适应算法。理论分析证明了该方法的有效性,在实际执行过程中精确可行。最后,提出了一种新的基于双麦克风阵列的噪声源跟踪方法,该方法使用两个相同的双麦克风阵列进行平面布站,继而采用经典快速的MUSIC算法获得各麦克风阵列提供的噪声源角度信息;通过扩展卡尔曼滤波对动态噪声源进行跟踪。仿真实验表明,无论动态噪声做简单直线运动还是变速曲线运动,该方法都具有良好的跟踪性能。更多还原

【Abstract】 Directionality technology is a key method solving hearing aids users’understanding difficulty under noise environment at present. It includes acoustic source localization and noise source tracking. The available directionality technology in China lacks of accuracy and real-time tracking ability. According to these problems, acoustic source localization methods based on microphones array and a noise tracking algorithm are presented in this paper. An adaptive directionality system is further illustrated.Firstly, this paper deals with time delay estimation algorithms (TDE). The improvements to two classical algorithms are also developed: In order to improve the estimation accuracy and convergence speed of the LMS adaptive filtering time delay estimation algorithm, we propose to utilize a combined filter transfer function to weight the phase data before fitting the phase data to the time delay estimation. For the adaptive eigenvalue decomposition (AED) time delay estimation algorithm, a new version combined with LMS time delay estimation is proposed, which converges more quickly to the new time delay value.Secondly, this paper researches an effective adaptive directionality system. This adaptive directionality system is based on an adaptive combination of two fixed polar patterns that act to make the null of the combined polar pattern of the system output always be toward the direction of the noise, and related adaptive algorithms to update this gain are also given. Theoretical analyses demonstrate its effectiveness and high accuracy.Finally, a novel and effective method for noise tracking based on two microphones array is proposed. This method uses two arrays disposed by 2-D locating to determine the position of noise exciter, and then uses the classic MUSIC algorithm to confirm the noise bearing information in real time. Extended Kalman Filter. (EKF) was employed to track the noise exciter using the above bearing estimation. Simulations show that this method is insensitive to the motions of the target and has the potential for practical applications.更多还原