The Relationship between Frequency Domain Blind Source Separation and Frequency Domain Adaptive Beamformer

Abstract:

As two important methods of array signal processing, blind source separation and beamforming can extract the target signal and suppress interference by using the received information of the array element. In the case of convolution mixture of sources, frequency domain blind source separation and frequency domain adaptive beamforming have similar signal model. To find the relationship between them, comparison between the minimization of the off-diagonal components in the BSS update equation and the minimization of the mean square error in the ABF had been made from the perspective of mathematical expressions, and find that the unmixing matrix of the BSS and the filter coefficients of the ABF converge to the same solution in the mean square error sense under the condition that the two source signals are ideally independent. With MATLAB, the equivalence in the frequency domain have been verified and the causes affecting separation performance have been analyzed, which was achieved by simulating instantaneous and convolution mixtures and separating mixture speech in frequency-domain blind source separation and frequency domain adaptive beamforming way.

• Periodical: Applied Mechanics and Materials (Volumes 490-491)
• Pages: 654-662
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.490-491.654
• Online since: January 2014
• Authors: Si Chong Qian*, Yang Xiang
• Keywords: Convolutive Mixtures, Frequency Domain Adaptive Beamformer, Frequency Domain Blind Source, Speech Separation

References

[1] R. B. Wang, C. H. Hou, J. Yang: The Relationship between Blind Separation of Instantaneous Mixture of Sources and Optimum Beamforming, Applied Acoustics. Vol. 30 (4): 302-305 (2011).

[2] Y. C. Wang, L. J. Chen, Wideband Blind Beamforming and Blind Separation of Convolved Mixtures. Vol. 22(1): 40-43 (2003).

[3] S. Kurita, H. Saruwatari, S. Kajita, K. Takeda, et al: Evaluation of Blind Signal Separation Method using Directivity Pattern under Reverberant Conditions/Proc. ICASSP2000.: 3140-3143 (2000).

DOI: 10.1109/icassp.2000.861203

[4] L. Parra and C. Alvino: Geometric Source Separation: Merging Convolutive Source Separation with Geometric Beamforming. Speech and Audio Processing, Vol 10(6): 352-362 (2001).

DOI: 10.1109/tsa.2002.803443

[5] L. Parra and C. Spence: Convolutive Blind Separation of Nonstationary Sources. IEEE Trans. Speech Audio Processing, Vol. 8(3): 320-327 (2000).

DOI: 10.1109/89.841214

[6] M. S. Pedersen, D. L. Wang, J. Larsen et al: Two-Microphone Separation of Speech Mixtures. IEEE transactions on neural networks, vol. 19 : 475-492 (2008).

DOI: 10.1109/tnn.2007.911740

[7] R. Prasad, H. Saruwatari, A. Lee et al: A Fixed Point ICA Algorithm for Convoluted Speech Separation. Proc. International Symposium on ICA＆BSS, Vol. 23(4): 579-584 (2003).

[8] Rajkishore Prasad, Hiroshi Saruwatari and Kiyohiro Shikano: Blind Separation of Speech by Fixed-point ICA with Source Adaptive Negentropy Approximation. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, E88-A(7): 1683-1692 (2005).

DOI: 10.1093/ietfec/e88-a.7.1683

[9] O. L. Frost: An Algorithm for Linearly Constrained Adaptive Array Processing. Proc. IEEE, Vol. 60(8): 926-935 (1972).

DOI: 10.1109/proc.1972.8817

[10] M. S. Brandstein and D. B. Ward: Microphone Arrays/Digital Signal Processing. New York: Springer-Verlag, (2001).