Array signal processing for audio signal separation and sound field reconstruction

In this paper, two innovation algorithms, which are the audio signal separation algorithm and the sound field reconstruction algorithm, are presented. The audio signal separation algorithm is based on the time-reversal method (TRM). The first step in this procedure is to calculate the impulse response function (IRF) between each source and field points by using deconvolution process. The second step in this procedure is to compute the passive time-reversal process on the signals of field points. Then, the signal of specific source can be separated via the self-adaptive focusing of time-reversal. The performance of TRM can be enhanced by increasing the number of sensors, the spacing of array sensor or the length of array and decreasing the measuring distance. The sound field reconstruction algorithm is based on the similar source method (SSM). The first step in this procedure is to search the source location by using beamforming approach. The second step in this procedure is to solve the source strength of virtual sources by using SSM. Finally, the sound pressure distribution of sound field can be reconstructed via the virtual sources, which have replaced the source. The results for simulation and experiment indicate following conclusions, the algorithm is not confined to a condition in which the measuring spacing must be smaller than wavelength of source. This algorithm can also reconstruct a sound field within measuring distance. Furthermore, the Tikhonov regularization process is used to avoid the singularity effect from noise in deconvolution process and inverse calculation which are included in the two algorithms, respectively. Finally, several simulations and experimentations are shown to verify two innovation algorithms in the present study.