Geometric Calibration of Distributed Microphone Arrays Based on Cross-Correlation Function

In this paper, we provide a solution to achieve the robustness of geometric calibration of distributed microphone arrays, under the condition that each microphone array has a non-ideal knowledge of the position of the microphones. We consider two microphone arrays located in arbitrary positions and propose a robust algorithm for sound source mapping based on cross-correlation function by placing a sound source in N different positions in the visual range of the two microphone arrays. An estimation is provided also on the rotation matrix R and translation vector t, which can represent the position and direction between two microphone arrays, using the techniques of Computer Vision. Experimental results demonstrate that the calibration algorithm based on cross-correlation function still performs well even in presence of a high microphone position error.