项目名称: 稳健的一阶和高阶谐波声场分析声源定位方法研究
项目编号: No.61471190
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 无线电电子学、电信技术
项目作者: 陈华伟
作者单位: 南京航空航天大学
项目金额: 83万元
中文摘要: 混响噪声和阵列失配误差是影响麦克风阵列声源定位性能的重要因素,如何实现混响噪声环境和阵列失配误差条件下的稳健声源定位,特别是基于小尺寸阵列的稳健声源定位,是音频与语音信号处理领域亟需解决的重要课题。一阶和高阶谐波声场分析为实现小尺寸阵列条件下的声源定位提供了重要技术途径。本项目针对混响噪声和麦克风失配误差的影响,系统研究基于一阶和高阶谐波声场分析的稳健声源定位的关键技术和方法,研究内容包括:基于一阶谐波声场分析的空域-时域-频域联合混响抑制和短时加权平均复声强估计的稳健声源定位方法;基于一阶谐波声场分析和机器学习理论的自校准稳健声源定位方法;基于高阶谐波声场分析的超指向性波束形成器的稳健性能分析与设计,以及具有低计算复杂度的高阶谐波声场分析可控波束形成稳健声源定位方法。本项目的研究对实现混响噪声环境和阵列失配条件下小尺寸阵列的稳健声源定位,具有重要的理论意义和实际应用价值。
中文关键词: 传声器阵列;声源定位算法;麦克风阵列
英文摘要: Reverberant noise and array mismatch errors are the key factors affecting the performance of sound localization using microphone arrays. How to achieve robust sound localization in reverberant environment with array mismatches, especially for small-sized microphone arrays, is a crucial issue in audio and speech processing field that needs to be addressed. The first- and higher-order harmonic sound field analysis has provided an important means to sound localization with small-sized microphone arrays. This project studies the key techniques and approaches of robust sound localization using first- and higher-order harmonic sound field analysis. The main contents include: the study of sound localization methods with the dereverberation in joint spatial-, temporal-, and spectral-domain and with the estimation of short-time weighted mean complex sound intensity using first-order harmonic sound field analysis; the study of sound localization methods with self-calibration capability based on first-order harmonic sound field analysis and machine learning theories; and the study of robust design of superdirective beamformer using higher-order harmonic sound field analysis and its robustness performance analysis, and the study of robust sound localization with lower computational complexity using the steered response power technique based on higher-order harmonic sound field analysis. This project is of important interest theoretically and practically to realize robust sound localization using small-sized microphone arrays in reverberant environments with sensor mismatches.
英文关键词: acoustic sensor arrays;sound source localization algorithms;microphone arrays