Fast deconvolution method for solid sphere array three-dimensional sound source recognition

Abstract

The invention discloses a fast deconvolution method for solid sphere array three-dimensional sound source recognition. The method comprises the following steps: 1, a sound pressure output formula is used for calculating the sound pressure output amount of each focusing grid point; 2, an output matrix is constructed by using the output amount of each point obtained through calculation; 3; a PSF function is used for calculating a PSF matrix A-tilde at the center focusing point; and 4, a sound source strength distribution matrix S-tilde is solved iteratively. The spatial shift immovability feature is used in the fourth step, a huge A matrix does not need to be calculated, a small-dimension matrix A-tilde only needs to be calculated, A<C>-tilde is obtained through A-tilde deformation, Fourier transform is carried out on the A<C>-tilde, based on FFT acceleration, the calculation time is greatly shortened, the calculation efficiency is improved, a good spatial resolution can be kept, and each sound source can be positioned accurately.

Description

technical field [0001] The invention relates to the technical field of sound source identification. Background technique [0002] Beamforming based on microphone array measurement has occupied an indispensable position in the field of sound source identification, and two-dimensional plane and three-dimensional solid sphere are commonly used array forms. The two-dimensional planar array is suitable for identifying the sound source in a specific opening angle area in front of the array, and the commonly used algorithm is delay summation (DAS); the three-dimensional solid spherical array can comprehensively identify the sound source in the three-dimensional chamber environment, and the commonly used algorithm is spherical harmonic Function Beamforming (SHB). Regardless of DAS or SHB, the output results can be regarded as the sum of the products of the intensity of each sound source and the corresponding point spread function (PSF function). In practical applications, factors s...

AI Technical Summary

Problems solved by technology

Regardless of DAS or SHB, the output results can be regarded as the sum of the products of the intensity of each sound source and the corresponding point spread function (PSF function). In practical applications, factors such as discrete sampling of the microphone make the PSF functions of the two algorithms equal It cannot be equal to the ideal delta function, not only outputs a wide "main lobe" at the real sound source position, but also outputs a high "side lobe" at the non-sound source position, which eventually leads to poor spatial resolution of the sound source identification result and low effective dynamic range

[0003] "Deconvolution for three-dimensional acoustic source identification based on spherical harmonics beamforming", Z.G.Chu, Y.Yang, Y.S.He. Journal of Sound and Vibration, Volume 344, Pages 484-502, 26May 2015 (three-dimensional deconvolution based on spherical harmonics beamforming Convolutional Sound Source Identification, Chu Zhigang, Yang Yang, He Yansong, Journal of Sound and Vibration, Volume 344, Pages 484-502, May 26, 2015) introduced a common solid sphere array 3D sound source identification deconvolution method , such as non-negative least squares (Non-Negative Least Squares, NNLS), NNLS can obtain clean and clear three-dimensional sound source imaging, but its calculation time is very long and the efficiency is very low, which cannot meet the needs of beamforming real-time imaging

Images