Underdetermined blind source separation based on continuous density hidden Markov models

Xiaoming Zhu; Keshab K Parhi

doi:10.1109/ICASSP.2010.5495730

Underdetermined blind source separation based on continuous density hidden Markov models

Xiaoming Zhu, Keshab K Parhi

Electrical and Computer Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1 Scopus citations

Abstract

In this paper, a novel method is developed to solve the problem of underdetermined blind source separation, where the number of mixtures is smaller than that of sources. Generalized Gaussian Distributions (GGDs) are used to model the source signals and generative Continuous Density Hidden Markov Models (CDHMMs) are derived to track the nonstationarity inside the source signals. Each source signal can switch between several states such that the separation performance can be significantly improved. The model parameters are trained through the Expectation Maximization (EM) algorithm and the source signals are estimated via the Maximum a Posteriori (MAP) approach. Compared with the results of L₁-norm solution, our proposed algorithm has obtained much better output signal-to-noise ratio (SNR) and the separation results are more realistic.

Original language	English (US)
Title of host publication	2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	4126-4129
Number of pages	4
ISBN (Print)	9781424442966
DOIs	https://doi.org/10.1109/ICASSP.2010.5495730
State	Published - 2010
Event	2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Dallas, TX, United States Duration: Mar 14 2010 → Mar 19 2010

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)	1520-6149

Other

Other	2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010
Country/Territory	United States
City	Dallas, TX
Period	3/14/10 → 3/19/10

Keywords

Generalized Gaussian distribution
Hidden Markov model
Nonstationary signals
Underdetermined blind source separation

Publisher link

10.1109/ICASSP.2010.5495730

Find It

Find It at U of M Libraries

Cite this

Zhu, X., & Parhi, K. K. (2010). Underdetermined blind source separation based on continuous density hidden Markov models. In 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings (pp. 4126-4129). Article 5495730 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2010.5495730

Underdetermined blind source separation based on continuous density hidden Markov models. / Zhu, Xiaoming; Parhi, Keshab K.
2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2010. p. 4126-4129 5495730 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Zhu, X & Parhi, KK 2010, Underdetermined blind source separation based on continuous density hidden Markov models. in 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings., 5495730, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, Institute of Electrical and Electronics Engineers Inc., pp. 4126-4129, 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010, Dallas, TX, United States, 3/14/10. https://doi.org/10.1109/ICASSP.2010.5495730

Zhu X, Parhi KK. Underdetermined blind source separation based on continuous density hidden Markov models. In 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2010. p. 4126-4129. 5495730. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP.2010.5495730

Zhu, Xiaoming ; Parhi, Keshab K. / Underdetermined blind source separation based on continuous density hidden Markov models. 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2010. pp. 4126-4129 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{624c54056b714dd6ad7c0bc143ecf3e4,

title = "Underdetermined blind source separation based on continuous density hidden Markov models",

abstract = "In this paper, a novel method is developed to solve the problem of underdetermined blind source separation, where the number of mixtures is smaller than that of sources. Generalized Gaussian Distributions (GGDs) are used to model the source signals and generative Continuous Density Hidden Markov Models (CDHMMs) are derived to track the nonstationarity inside the source signals. Each source signal can switch between several states such that the separation performance can be significantly improved. The model parameters are trained through the Expectation Maximization (EM) algorithm and the source signals are estimated via the Maximum a Posteriori (MAP) approach. Compared with the results of L1-norm solution, our proposed algorithm has obtained much better output signal-to-noise ratio (SNR) and the separation results are more realistic.",

keywords = "Generalized Gaussian distribution, Hidden Markov model, Nonstationary signals, Underdetermined blind source separation",

author = "Xiaoming Zhu and Parhi, {Keshab K}",

year = "2010",

doi = "10.1109/ICASSP.2010.5495730",

language = "English (US)",

isbn = "9781424442966",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "4126--4129",

booktitle = "2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings",

note = "2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 ; Conference date: 14-03-2010 Through 19-03-2010",

}

TY - GEN

T1 - Underdetermined blind source separation based on continuous density hidden Markov models

AU - Zhu, Xiaoming

AU - Parhi, Keshab K

PY - 2010

Y1 - 2010

N2 - In this paper, a novel method is developed to solve the problem of underdetermined blind source separation, where the number of mixtures is smaller than that of sources. Generalized Gaussian Distributions (GGDs) are used to model the source signals and generative Continuous Density Hidden Markov Models (CDHMMs) are derived to track the nonstationarity inside the source signals. Each source signal can switch between several states such that the separation performance can be significantly improved. The model parameters are trained through the Expectation Maximization (EM) algorithm and the source signals are estimated via the Maximum a Posteriori (MAP) approach. Compared with the results of L1-norm solution, our proposed algorithm has obtained much better output signal-to-noise ratio (SNR) and the separation results are more realistic.

AB - In this paper, a novel method is developed to solve the problem of underdetermined blind source separation, where the number of mixtures is smaller than that of sources. Generalized Gaussian Distributions (GGDs) are used to model the source signals and generative Continuous Density Hidden Markov Models (CDHMMs) are derived to track the nonstationarity inside the source signals. Each source signal can switch between several states such that the separation performance can be significantly improved. The model parameters are trained through the Expectation Maximization (EM) algorithm and the source signals are estimated via the Maximum a Posteriori (MAP) approach. Compared with the results of L1-norm solution, our proposed algorithm has obtained much better output signal-to-noise ratio (SNR) and the separation results are more realistic.

KW - Generalized Gaussian distribution

KW - Hidden Markov model

KW - Nonstationary signals

KW - Underdetermined blind source separation

UR - http://www.scopus.com/inward/record.url?scp=78049392547&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78049392547&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2010.5495730

DO - 10.1109/ICASSP.2010.5495730

M3 - Conference contribution

AN - SCOPUS:78049392547

SN - 9781424442966

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 4126

EP - 4129

BT - 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2010

Y2 - 14 March 2010 through 19 March 2010

ER -

Underdetermined blind source separation based on continuous density hidden Markov models

Abstract

Publication series

Other

Keywords

Publisher link

Find It

Other files and links

Fingerprint

Cite this