基于矩形麦克风阵列的改进的GCC-PHAT语音定位算法

摘要

针对相位变换加权广义互相关方法（GCC-PHAT）对噪声的影响较为敏感的缺点，本文通过削弱噪音互谱、加权信噪比、应用相干函数等手段对原始的相位变换加权函数（PHAT）进行了改进，得到了一种改进的相位变换加权函数（MPHAT），以便在有噪情况下更准确地估计时间差。大量的仿真实验验证了本文算法的有效性。%This paper improves a phase transform weighting function （PHAT） by weakening the noise cross-spectrum, weighting signal-to-noise ratio and applying the coherence function in view of the negative influence of the sensitivity of generalized cross correlation-phase transform （GCC-PHAT） algorithm to noise effect. We therefore acquired an improved phase transform weighting function （IPHAT） to facilitate the more accurate estimation of time arrival difference in the effect of noise. Vast experiments prove the effectiveness of the presented algorithm.