Model-based Noise PSD Estimation from Speech in Non-stationary Noise

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

7 Citations (Scopus)
550 Downloads (Pure)

Abstract

Most speech enhancement algorithms need an estimate of the noise power spectral density (PSD) to work. In this paper, we introduce a model-based framework for doing noise PSD estimation. The proposed framework allows us to include prior spectral information about the speech and noise sources, can be configured to have zero tracking delay, and does not depend on estimated speech presence probabilities. This is in contrast to other noise PSD estimators which often have a too large tracking delay to give good results in non- stationary situations and offer no consistent way of including prior information about the speech or the noise type. The results show that the proposed method outperforms state-of-the-art noise PSD estima- tors in terms of tracking speed and estimation accuracy.
Original languageEnglish
Title of host publicationIEEE International Conference on Acoustics, Speech, and Signal Processing
Number of pages5
Place of PublicationCalgary, Canada
PublisherIEEE
Publication date10 Sep 2018
Pages5424-5428
Article number8461683
ISBN (Electronic)978-1-5386-4658-8
DOIs
Publication statusPublished - 10 Sep 2018
Event2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) - Calgary, Canada
Duration: 15 Apr 201820 Apr 2018
https://2018.ieeeicassp.org/

Conference

Conference2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
CountryCanada
CityCalgary
Period15/04/201820/04/2018
Internet address
SeriesI E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings
ISSN1520-6149

Fingerprint

Power spectral density
Speech enhancement

Keywords

  • Noise PSD estimation
  • Noise statistics
  • Speech enhancement

Cite this

Nielsen, J. K., Kavalekalam, M. S., Christensen, M. G., & Boldt, J. B. (2018). Model-based Noise PSD Estimation from Speech in Non-stationary Noise. In IEEE International Conference on Acoustics, Speech, and Signal Processing (pp. 5424-5428). [8461683] Calgary, Canada: IEEE. I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings https://doi.org/10.1109/ICASSP.2018.8461683
Nielsen, Jesper Kjær ; Kavalekalam, Mathew Shaji ; Christensen, Mads Græsbøll ; Boldt, Jesper Bünsow. / Model-based Noise PSD Estimation from Speech in Non-stationary Noise. IEEE International Conference on Acoustics, Speech, and Signal Processing. Calgary, Canada : IEEE, 2018. pp. 5424-5428 (I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings).
@inproceedings{05b412dd0a7e4b319b3dcff8625881bb,
title = "Model-based Noise PSD Estimation from Speech in Non-stationary Noise",
abstract = "Most speech enhancement algorithms need an estimate of the noise power spectral density (PSD) to work. In this paper, we introduce a model-based framework for doing noise PSD estimation. The proposed framework allows us to include prior spectral information about the speech and noise sources, can be configured to have zero tracking delay, and does not depend on estimated speech presence probabilities. This is in contrast to other noise PSD estimators which often have a too large tracking delay to give good results in non- stationary situations and offer no consistent way of including prior information about the speech or the noise type. The results show that the proposed method outperforms state-of-the-art noise PSD estima- tors in terms of tracking speed and estimation accuracy.",
keywords = "Noise PSD estimation, Noise statistics, Speech enhancement",
author = "Nielsen, {Jesper Kj{\ae}r} and Kavalekalam, {Mathew Shaji} and Christensen, {Mads Gr{\ae}sb{\o}ll} and Boldt, {Jesper B{\"u}nsow}",
year = "2018",
month = "9",
day = "10",
doi = "10.1109/ICASSP.2018.8461683",
language = "English",
series = "I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings",
publisher = "IEEE",
pages = "5424--5428",
booktitle = "IEEE International Conference on Acoustics, Speech, and Signal Processing",
address = "United States",

}

Nielsen, JK, Kavalekalam, MS, Christensen, MG & Boldt, JB 2018, Model-based Noise PSD Estimation from Speech in Non-stationary Noise. in IEEE International Conference on Acoustics, Speech, and Signal Processing., 8461683, IEEE, Calgary, Canada, I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings, pp. 5424-5428, 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Calgary, Canada, 15/04/2018. https://doi.org/10.1109/ICASSP.2018.8461683

Model-based Noise PSD Estimation from Speech in Non-stationary Noise. / Nielsen, Jesper Kjær; Kavalekalam, Mathew Shaji; Christensen, Mads Græsbøll; Boldt, Jesper Bünsow.

IEEE International Conference on Acoustics, Speech, and Signal Processing. Calgary, Canada : IEEE, 2018. p. 5424-5428 8461683 (I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings).

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Model-based Noise PSD Estimation from Speech in Non-stationary Noise

AU - Nielsen, Jesper Kjær

AU - Kavalekalam, Mathew Shaji

AU - Christensen, Mads Græsbøll

AU - Boldt, Jesper Bünsow

PY - 2018/9/10

Y1 - 2018/9/10

N2 - Most speech enhancement algorithms need an estimate of the noise power spectral density (PSD) to work. In this paper, we introduce a model-based framework for doing noise PSD estimation. The proposed framework allows us to include prior spectral information about the speech and noise sources, can be configured to have zero tracking delay, and does not depend on estimated speech presence probabilities. This is in contrast to other noise PSD estimators which often have a too large tracking delay to give good results in non- stationary situations and offer no consistent way of including prior information about the speech or the noise type. The results show that the proposed method outperforms state-of-the-art noise PSD estima- tors in terms of tracking speed and estimation accuracy.

AB - Most speech enhancement algorithms need an estimate of the noise power spectral density (PSD) to work. In this paper, we introduce a model-based framework for doing noise PSD estimation. The proposed framework allows us to include prior spectral information about the speech and noise sources, can be configured to have zero tracking delay, and does not depend on estimated speech presence probabilities. This is in contrast to other noise PSD estimators which often have a too large tracking delay to give good results in non- stationary situations and offer no consistent way of including prior information about the speech or the noise type. The results show that the proposed method outperforms state-of-the-art noise PSD estima- tors in terms of tracking speed and estimation accuracy.

KW - Noise PSD estimation

KW - Noise statistics

KW - Speech enhancement

UR - http://www.scopus.com/inward/record.url?scp=85054270558&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2018.8461683

DO - 10.1109/ICASSP.2018.8461683

M3 - Article in proceeding

T3 - I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings

SP - 5424

EP - 5428

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

PB - IEEE

CY - Calgary, Canada

ER -

Nielsen JK, Kavalekalam MS, Christensen MG, Boldt JB. Model-based Noise PSD Estimation from Speech in Non-stationary Noise. In IEEE International Conference on Acoustics, Speech, and Signal Processing. Calgary, Canada: IEEE. 2018. p. 5424-5428. 8461683. (I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings). https://doi.org/10.1109/ICASSP.2018.8461683