复杂环境下基于时延估计的声源定位技术研究

发布时间：2018-06-12 04:41

本文选题：时延估计 + 传递函数比　；参考：《通信学报》2014年01期

【摘要】：为了改善在复杂环境下声源定位算法的性能,提出了一种新的时延估计(TDE)方法,即基于传递函数比的统计模型方法(ATFR-SM)。该方法采用统计模型去除噪声对传递函数(ATF)的影响,在计算传递函数时对功率谱密度(PSD)进行平滑和"白化",以去除混响对传递函数的影响。同时,算法中引入话音激活检测(VAD)去除对求取传递函数无用的噪声段,以提高时延估计的准确性。此外,将所提时延估计方法与线性定位法相结合,构成一套完整的声源定位方法。实验结果表明,在复杂环境下,时延估计方法具有更低的异常点百分比(PAP)和均方根误差(RMSE),且明显优于传统的参考算法,同时声源定位方法具有更高的定位精度。
[Abstract]:In order to improve the performance of acoustic source location algorithm in complex environments, a new time delay estimation (TDE) method is proposed, which is a statistical model method based on transfer function ratio. The statistical model is used to remove the effect of noise on the transfer function (ATF), and the power spectral density (PSD) is smoothed and "whitened" when calculating the transfer function to remove the effect of reverberation on the transfer function. In order to improve the accuracy of time delay estimation, voice activation detection (VAD) is introduced into the algorithm to remove the noise segments that are useless for obtaining the transfer function. In addition, the proposed time delay estimation method is combined with the linear location method to form a complete sound source location method. The experimental results show that the time delay estimation method has a lower percentage of outliers (PAPs) and root mean square error (RMSE) in complex environments, and it is obviously superior to the traditional reference algorithm. At the same time, the acoustic source localization method has higher localization accuracy.
【作者单位】：北京工业大学电子信息与控制工程学院
【基金】：北京市教育委员会科技发展计划重点基金资助项目(KZ201110005005) 国家自然科学基金资助项目(61072089)~~
【分类号】：TN912.3

【参考文献】