Research On The Robustness Of Nearfield Microphone Array Algorithms

The directivity of microphone array has been intensely investigated in acoustic signal processing field. The normally used array algorithms are based on the far field acoustic model, which is not fit for many application scenarios, including small meeting rooms and car cabins. On the other side, with the widespread use of portable communication devices, more and more attention has been paid to the nearfield signal acquisition. This also demands that the array should have a high nearfield focusing performance. It should be noted that the nearfield beamformers are more vulnerable to the sound field variation. In this thesis, the robustness of nearfield microphone array algorithms is investigated.First of all, two highly robust beamformer strategies are picked out by comparison of different nearfield array algorithms. The first strategy is the maximization of the white noise gain of the array, and the delay and sum algorithm is the standard algorithm using this strategy. The second strategy is to introduce the sound field variation model into the optimization of the array parameters, and the robust focusing method is a typical algorithm using this strategy.The room acoustic model is used to analyze the variation of the microphone response brought by the reverberation. The influence of different variation distribution models on the performance of robust focusing method is investigated, based on which the optimized variation parameters for robust focusing method are proposed. Several typical nearfield robust beamformers are compared under different room acoustic models.Finally, the experiments are carried out to validate the basic performance of robust nearfield beamformers. Two kinds of microphone arrays are used. One is a10-element line array, and the other is a32-element rectangular array.