Relaxing the WDO Assumption in Blind Extraction of Speakers from Speech Mixtures
DOI:
https://doi.org/10.26636/jtit.2010.4.1096Keywords:
blind source extraction, harmonic frequencies, histogram clustering, spectrogram analysis, speech reconstruction, time-frequency masking, W-disjoint orthogonalAbstract
The time-frequency masking approach in blind speech extraction consists of two main steps: feature clustering in a space spanned over delay-time and attenuation rate, and spectrogram masking in order to reconstruct the sources. Usually a binary mask is generated under the strong W-disjoint orthogonal (WDO) assumption (disjoint orthogonal representations in the frequency domain). In practice, this assumption is most often violated leading to weak quality of reconstructed sources. In this paper we propose the WDO to be relaxed by allowing some frequency bins to be shared by both sources. As we detect instantaneous fundamental frequencies the mask creation is supported by exploring a harmonic structure of speech. The proposed method is proved to be effective and reliable in experiments with both simulated and real acquired mixtures.
Downloads
References
[1] S. Makino, T.-W. Lee, and H. Sawada, Eds., Blind Speech Separation. Berlin: Springer, 2007.
View in Google Scholar
[2] A. Hyvärinen, J. Karhunen, and E. Oja, Independent Component Analysis. New York: Wiley, 2001.
View in Google Scholar
[3] A. Cichocki and S. Amari, Adaptive Blind Signal and Image Processing. Chichester: Wiley, 2003.
View in Google Scholar
[4] O. Yilmaz and S. Rickard, „Blind separation of speech mixtures via time-frequency masking”, IEEE Trans. Sig. Proces., vol. 52, no. 7, pp. 1830-1847, 2004.
View in Google Scholar
[5] S. Rickard, „The DUET blind source separation algorithm”, in [1], pp. 217-237.
View in Google Scholar
[6] M. Aoki et al., „Sound source segregation based on estimating incident angle of each frequency component of input signals acquired by multiple microphones”, Acoust. Sci. Technol., vol. 22, no. 2, pp. 149-157, 2001.
View in Google Scholar
[7] F. Abrard and Y. Deville, „A time-frequency blind signal separation method applicable to underdetermined mixtures of dependent sources”, Sig. Proces., vol. 85, pp. 1389-1403, 2005.
View in Google Scholar
[8] S. Arberet, R. Gribonval, and F. Bimbot, „A robust method to count, locate and separate audio sources in a multichannel underdetermined mixture”, IEEE Trans. Sig. Proces., vol. 58, no. 1, pp. 121-133, 2010.
View in Google Scholar
[9] Z. He, A. Cichocki, Y. Li, S. Xie, and S. Sanei, „K-hyperline clustering learning for sparse component analysis”, Sig. Proces., vol. 89, pp. 1011-1022, 2009.
View in Google Scholar
[10] S. Makino, H. Sawada, R. Mukai, and S. Araki, „Blind source separation of convolutive mixture of speech in frequency domain”, IEICE Trans. Fundament., vol. 88, no. 7, pp. 1830-1847, 2004.
View in Google Scholar
[11] S. Araki, H. Sawada, R. Mukai, and S. Makino, „Underdetermined blind sparse source separation for arbitrarily arranged multiple sensors”, Sig. Proces., vol. 87, pp. 1833-1847, 2007.
View in Google Scholar
[12] H. Ouchi and N. Hamada, „Separation of speech mixture by time-frequency masking utilizing sound harmonics”, J. Sig. Proces., vol. 13, no. 4, pp. 331-334, 2009.
View in Google Scholar
[13] T. Kobayasi, S. Itahashi, S. Hayamizu, and T. Takezawa, „ASJ continuous speech corpus for research”, J. Acoust. Soc. Japan, vol. 48, no. 12, pp. 888-893, 1992 (in Japanese).
View in Google Scholar
Downloads
Submitted
Published
Issue
Section
License
Copyright (c) 2023 Journal of Telecommunications and Information Technology
This work is licensed under a Creative Commons Attribution 4.0 International License.
