基于迭代结构的双耳声源定位算法研究
毕业论文关键词 定位 迭代 耳间时间差
Title Sound Source Localization based on Iteration Method Abstract Sound source localization is fundamental in speech signal processing technique. Binaural solution becomes difficult when facing noise and reverberation. This thesis improves conventional localization method based on ITD (Interaural Time Difference) estimation and proposes an iteration method for localization. Conventional ITD values from classic GCC method differs little between adjacent source directions，which is easily causing mistakes. This thesis employs cubic spline interpolation in calculating cross correlation function and obtains more ideal ITD estimated values, which perform better in simulation experiments. In this thesis we propose an iteration method for localization, which combines speech separation and source localization. Simply speaking, we firstly estimate locations of multiple sources from mixed speech, and then separate the mixed speech according to the sources` locations; secondly we re-estimate source locations from the separated speeches.
Keywords localization iteration Interaural Time Difference
目录
第一章绪论1
1.1双耳声源定位技术的研究背景和意义1
1.2声源定位技术的研究现状..1
1.3论文主要研究内容..2
1.4论文组织结构.2
第二章双耳声源定位理论5
2.1人类听觉基础.5
2.1.1人耳听觉的基础结构..5
2.1.2空间听觉.5
2.2双耳声源定位线索..7
2.2.1耳间时间差7
2.2.2耳间强度差8
2.2.3双耳互相关函数.9
2.3双耳语音信号的采集..10
2.3.1与头相关传递函数.10
2.3.2双耳信号的计算机模拟..11
2.3.3人工头采集双耳信号12
2.4基于双耳空间定位的语音分离概述..13
2.4.1原有的语音定位分离算法.13
2.4.2迭代的语音分离算法14
2.4.3实验与性能评估算法15
2.5本章小结.16
第三章基于ITD的双耳声源定位17
3.1双耳空间线索提取17
3.1.1预处理17
3.1.2特征提取..18
3.2基于ITD的声源定位原理..19
3.3对ITD定位方法的改进21
3.4定位效果的评估.24
3.5本章小结.25
第四章定位与分离的迭代方法..27
4.1迭代结构.27
4.2混合语音的初步定位..27
4.3分离方法.27
4.4重定位与重分离方法..29
4.5评估结果.29
4.6本章小结.31
第五章总结与展望.33