摘要
声源定位技术是语音增强、语音识别技术的前提和基础。基于麦克风阵列的声源定位技术已经成为一大研究热点,其广阔的应用前景得到了广泛的关注。本文提出基于变步长标准最小均方差VLMS(Variable Step Size Least Mean Square)的声源定位算法。该算法利用VLMS算法自适应估计声源到麦克风的脉冲响应系数,进而估计出各麦克风之间时延,并利用几何方法定位声源在3D空间的位置。此外,本文设计了基于Cortex-A8嵌入式平台的声源定位系统,并进行了相应的硬件选型与调试及算法移植工作。实时实验显示,本系统的方案合理有效,能够较好的实现声源定位。
Sound localization technology is the precondition and foundation of speech enhancement and speech recognition technology. Sound localization technology based on microphone array has become a hot research topic,and its broad application prospect has been widely concerned. A real sound localization method based on the variable step size least mean square（VLMS）is proposed. The proposed method estimates the impulse response coefficients between the speech source and microphones by means of adaptive subgradient projection algorithm,then acquires the time delays of microphone pairs,and calculates the source position in 3D space by geometric method subsequently. Results showed that the algorithm has low computing complexity and strong practicability. In addition,the sound localization system based on Cortex-A8 architecture embedded platform is designed. Real experiments show that the scheme of this system is reasonable and effective,and can locate the sound source well.
出处
《电子器件》
CAS
北大核心
2015年第6期1406-1411,共6页
Chinese Journal of Electron Devices
基金
国家自然基金项目(61375028)
江苏省自然基金项目(BK20130241)
泉州市科技计划项目(2014Z103)
关键词
麦克风阵列
声源定位
四元十字阵
时延估计
microphone array
sound location
four-element cross array
time delay estimation
