Experimental Analysis and Design Guidelines for Microphone Virtualization in Automotive Scenarios

In this paper, a performance analysis on the estimation of so-called observation filter for Virtual Microphone Technique (VMT) in realistic automotive environment is presented. A comparison between adaptive and fixed methods, namely Least Mean Square (LMS) Minimum Error (MMSE), respectively, was carried on. Two different experimental setups were implemented popular B-segment car. Eight microphones placed at monitoring virtual positions order to sense environmental acoustic noise propagating within cabin car running variable speed smooth asphalt. Our results show that large spectral coherence microphone signals indicates potentially effective relatively wide-band signal reconstruction. The method achieves better than one, guaranteeing remarkable broadband accuracy. Moreover, each considered setup, design guidelines are proposed obtain good trade-off accuracy material costs.