Abstract:
Microphone arrays represent an attractive sound pickup solution for virtual acoustics applications where a high rejection of the acoustic environment (ambient noise and reverberation) is needed for minimizing artifacts during the postprocessing of the sound (filtering, spatialization, artificial reverberation). This study intends to evaluate, on the basis of simulations and experiments, different prototypes of microphone arrays designed at CNET. The different steps of the array processing (i.e., band filtering of subarrays, sensors weighting, beamforming, and steering) are implemented on a signal processing prototyping environment. This program is used to synthesize the impulse response of the microphone array in a real acoustic field. The input signals are the impulse responses successively measured between a source and each sensor location of the array in real rooms. The objective characterization of the array is based on classical criteria used in room acoustics and on the analysis of the time--frequency envelope of the late reverberation. The perceptual analysis is performed by convolution of the array responses with anechoic signals. A parametrical study is undertaken in order to study the influence of the geometrical and signal processing characteristics of the array. Theoretical and experimental results are compared and discussed. [Work supported by CNET.]