Frequency Response Calibration Using Multi-Channel Wiener Filters for Microphone Arrays

Frequency response differences among microphones will result in performance degradation of microphone arrays. To remedy this problem, a frequency response calibration method based on multi-channel Wiener filters (MCWFs) is proposed in this paper. The geometric relationship of a single sound source and uncalibrated microphones is first formulated via maximum allowable errors of gain and phase. Next, the system model of the frequency response calibration is constructed, and the microphone signals are approximated to each other using the MCWF. Finally, a least-mean-square (LMS) algorithm is presented to obtain the Wiener solutions rapidly. The proposed method can effectively compensate both the gain and phase errors caused by different sensitivities among microphones in noisy and reverberant environments. The simulation and real-word experimental results reveal the validity of the proposed method.