中文摘要：

针对基本互相关法在估计时延时没有明显的峰值及实际环境中的噪声会减损峰值，甚至出现伪峰的问题，提出了一种基于广义互相关的改进时延估计方法，并结合麦克风阵列结构，实现了基于两步定位的声源定位．通过仿真实验验证了该方法在有噪声的环境下能准确测量声源的位置，定位的平均距离误差和平均方向角误差都低于5％，且方向角在30°～150°、距离在1．1～2．3m时定位成功率达80％以上．

英文摘要：

In order to solve the problems that there would not appear obvious peak with the basic cross-correlation （BCC） in the time delay estimation （TDE） and the noise and reverberation in the actual environment would minify the BCC function＇s peak and even leads to fake peaks, an improved time delay estimation method based on the generalized cross-correlation （GCC） was presented and combined with the microphone array structure, so the two-step sound source localization method was implemented. Simulation results indicated that the sound source＇s location could be measured accurately with the new method in noised environment. The average errors in positioning distance and angle are all below 5%. The successful rate of positioning is above 80% with the direction anzle between 30° and 150° and the distance from 1.1 m to 2.3 m.