An inverse-gamma source variance prior with factorized parameterization for audio source separation

Dionyssos Kounades-Bastian; Laurent Girin; Xavier Alameda-Pineda; Sharon Gannot; Radu Horaud

doi:10.1109/ICASSP.2016.7471652

An inverse-gamma source variance prior with factorized parameterization for audio source separation

Dionyssos Kounades-Bastian, Laurent Girin, Xavier Alameda-Pineda, Sharon Gannot, Radu Horaud

Bar-Ilan University - The Alexander Kofkin Faculty of Engineering

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

6 Scopus citations

Abstract

In this paper we present a new statistical model for the power spectral density (PSD) of an audio signal and its application to multichannel audio source separation (MASS). The source signal is modeled with the local Gaussian model (LGM) and we propose to model its variance with an inverse-Gamma distribution, whose scale parameter is factorized as a rank-1 model. We discuss the interest of this approach and evaluate it in a MASS task with underdetermined convolutive mixtures. For this aim, we derive a variational EM algorithm for parameter estimation and source inference. The proposed model shows a benefit in source separation performance compared to a state-of-the-art LGM NMF-based technique.

Original language	English
Title of host publication	2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	136-140
Number of pages	5
ISBN (Electronic)	9781479999880
DOIs	https://doi.org/10.1109/ICASSP.2016.7471652
State	Published - 18 May 2016
Event	41st IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Shanghai, China Duration: 20 Mar 2016 → 25 Mar 2016

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	2016-May
ISSN (Print)	1520-6149

Conference

Conference	41st IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016
Country/Territory	China
City	Shanghai
Period	20/03/16 → 25/03/16

Bibliographical note

Publisher Copyright:
© 2016 IEEE.

Keywords

Audio modeling
PSD model
audio source separation
local Gaussian model

Access to Document

10.1109/ICASSP.2016.7471652

Cite this

Kounades-Bastian, D., Girin, L., Alameda-Pineda, X., Gannot, S., & Horaud, R. (2016). An inverse-gamma source variance prior with factorized parameterization for audio source separation. In 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings (pp. 136-140). [7471652] (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2016-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2016.7471652

Kounades-Bastian, Dionyssos ; Girin, Laurent ; Alameda-Pineda, Xavier et al. / An inverse-gamma source variance prior with factorized parameterization for audio source separation. 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 136-140 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{f751b73903fb4adfbbbc9a73b853e5f8,

title = "An inverse-gamma source variance prior with factorized parameterization for audio source separation",

abstract = "In this paper we present a new statistical model for the power spectral density (PSD) of an audio signal and its application to multichannel audio source separation (MASS). The source signal is modeled with the local Gaussian model (LGM) and we propose to model its variance with an inverse-Gamma distribution, whose scale parameter is factorized as a rank-1 model. We discuss the interest of this approach and evaluate it in a MASS task with underdetermined convolutive mixtures. For this aim, we derive a variational EM algorithm for parameter estimation and source inference. The proposed model shows a benefit in source separation performance compared to a state-of-the-art LGM NMF-based technique.",

keywords = "Audio modeling, PSD model, audio source separation, local Gaussian model",

author = "Dionyssos Kounades-Bastian and Laurent Girin and Xavier Alameda-Pineda and Sharon Gannot and Radu Horaud",

note = "Publisher Copyright: {\textcopyright} 2016 IEEE.; 41st IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 ; Conference date: 20-03-2016 Through 25-03-2016",

year = "2016",

month = may,

day = "18",

doi = "10.1109/ICASSP.2016.7471652",

language = "אנגלית",

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

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

pages = "136--140",

booktitle = "2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings",

address = "ארצות הברית",

}

Kounades-Bastian, D, Girin, L, Alameda-Pineda, X, Gannot, S & Horaud, R 2016, An inverse-gamma source variance prior with factorized parameterization for audio source separation. in 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings., 7471652, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2016-May, Institute of Electrical and Electronics Engineers Inc., pp. 136-140, 41st IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016, Shanghai, China, 20/03/16. https://doi.org/10.1109/ICASSP.2016.7471652

An inverse-gamma source variance prior with factorized parameterization for audio source separation. / Kounades-Bastian, Dionyssos; Girin, Laurent; Alameda-Pineda, Xavier et al.
2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2016. p. 136-140 7471652 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2016-May).

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

TY - GEN

T1 - An inverse-gamma source variance prior with factorized parameterization for audio source separation

AU - Kounades-Bastian, Dionyssos

AU - Girin, Laurent

AU - Alameda-Pineda, Xavier

AU - Gannot, Sharon

AU - Horaud, Radu

PY - 2016/5/18

Y1 - 2016/5/18

N2 - In this paper we present a new statistical model for the power spectral density (PSD) of an audio signal and its application to multichannel audio source separation (MASS). The source signal is modeled with the local Gaussian model (LGM) and we propose to model its variance with an inverse-Gamma distribution, whose scale parameter is factorized as a rank-1 model. We discuss the interest of this approach and evaluate it in a MASS task with underdetermined convolutive mixtures. For this aim, we derive a variational EM algorithm for parameter estimation and source inference. The proposed model shows a benefit in source separation performance compared to a state-of-the-art LGM NMF-based technique.

AB - In this paper we present a new statistical model for the power spectral density (PSD) of an audio signal and its application to multichannel audio source separation (MASS). The source signal is modeled with the local Gaussian model (LGM) and we propose to model its variance with an inverse-Gamma distribution, whose scale parameter is factorized as a rank-1 model. We discuss the interest of this approach and evaluate it in a MASS task with underdetermined convolutive mixtures. For this aim, we derive a variational EM algorithm for parameter estimation and source inference. The proposed model shows a benefit in source separation performance compared to a state-of-the-art LGM NMF-based technique.

KW - Audio modeling

KW - PSD model

KW - audio source separation

KW - local Gaussian model

UR - http://www.scopus.com/inward/record.url?scp=84973307938&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2016.7471652

DO - 10.1109/ICASSP.2016.7471652

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:84973307938

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

SP - 136

EP - 140

BT - 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 41st IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016

Y2 - 20 March 2016 through 25 March 2016

ER -

Kounades-Bastian D, Girin L, Alameda-Pineda X, Gannot S, Horaud R. An inverse-gamma source variance prior with factorized parameterization for audio source separation. In 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2016. p. 136-140. 7471652. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP.2016.7471652

An inverse-gamma source variance prior with factorized parameterization for audio source separation

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this