A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices

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

1 Citation (Scopus)
33 Downloads (Pure)

Abstract

The presence of background noise in signals adversely affects the performance of many speech-based algorithms. Accurate estimation of signal-to-noise-ratio (SNR), as a measure of noise level in a signal, can help in compensating for noise effects. Most existing SNR estimation methods have been developed for normal speech and might not provide accurate estimation for special speech types such as whispered or disordered voices, particularly, when they are corrupted by non-stationary noises. In this paper, we first investigate the impact of stationary and non-stationary noise on the behavior of mel-frequency cepstral coefficients (MFCCs) extracted from normal, whispered and pathological voices. We demonstrate that, regardless of the speech type, the mean and the covariance of MFCCs are predictably modified by additive noise and the amount of change is related to the noise level. Then, we propose a new supervised method for SNR estimation which is based on a regression model trained on MFCCs of the noisy signals. Experimental results show that the proposed approach provides accurate estimation and consistent performance for various speech types under different noise conditions.
Original languageEnglish
Title of host publication2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Number of pages5
PublisherIEEE
Publication date10 Sep 2018
Pages296-300
Article number8462459
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

Signal to noise ratio
Additive noise

Keywords

  • Global SNR estimation
  • MFCC
  • Pathological voice
  • Support vector regression
  • Whispered speech

Cite this

Poorjam, A. H., Little, M. A., Jensen, J. R., & Christensen, M. G. (2018). A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 296-300). [8462459] IEEE. I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings https://doi.org/10.1109/ICASSP.2018.8462459
Poorjam, Amir Hossein ; Little, Max A ; Jensen, Jesper Rindom ; Christensen, Mads Græsbøll. / A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices. 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018. pp. 296-300 (I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings).
@inproceedings{e0e5d796b29b4950b7f948b8042f6660,
title = "A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices",
abstract = "The presence of background noise in signals adversely affects the performance of many speech-based algorithms. Accurate estimation of signal-to-noise-ratio (SNR), as a measure of noise level in a signal, can help in compensating for noise effects. Most existing SNR estimation methods have been developed for normal speech and might not provide accurate estimation for special speech types such as whispered or disordered voices, particularly, when they are corrupted by non-stationary noises. In this paper, we first investigate the impact of stationary and non-stationary noise on the behavior of mel-frequency cepstral coefficients (MFCCs) extracted from normal, whispered and pathological voices. We demonstrate that, regardless of the speech type, the mean and the covariance of MFCCs are predictably modified by additive noise and the amount of change is related to the noise level. Then, we propose a new supervised method for SNR estimation which is based on a regression model trained on MFCCs of the noisy signals. Experimental results show that the proposed approach provides accurate estimation and consistent performance for various speech types under different noise conditions.",
keywords = "Global SNR estimation, MFCC, Pathological voice, Support vector regression, Whispered speech",
author = "Poorjam, {Amir Hossein} and Little, {Max A} and Jensen, {Jesper Rindom} and Christensen, {Mads Gr{\ae}sb{\o}ll}",
year = "2018",
month = "9",
day = "10",
doi = "10.1109/ICASSP.2018.8462459",
language = "English",
pages = "296--300",
booktitle = "2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)",
publisher = "IEEE",
address = "United States",

}

Poorjam, AH, Little, MA, Jensen, JR & Christensen, MG 2018, A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices. in 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)., 8462459, IEEE, I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings, pp. 296-300, 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Calgary, Canada, 15/04/2018. https://doi.org/10.1109/ICASSP.2018.8462459

A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices. / Poorjam, Amir Hossein; Little, Max A; Jensen, Jesper Rindom; Christensen, Mads Græsbøll.

2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018. p. 296-300 8462459.

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

TY - GEN

T1 - A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices

AU - Poorjam, Amir Hossein

AU - Little, Max A

AU - Jensen, Jesper Rindom

AU - Christensen, Mads Græsbøll

PY - 2018/9/10

Y1 - 2018/9/10

N2 - The presence of background noise in signals adversely affects the performance of many speech-based algorithms. Accurate estimation of signal-to-noise-ratio (SNR), as a measure of noise level in a signal, can help in compensating for noise effects. Most existing SNR estimation methods have been developed for normal speech and might not provide accurate estimation for special speech types such as whispered or disordered voices, particularly, when they are corrupted by non-stationary noises. In this paper, we first investigate the impact of stationary and non-stationary noise on the behavior of mel-frequency cepstral coefficients (MFCCs) extracted from normal, whispered and pathological voices. We demonstrate that, regardless of the speech type, the mean and the covariance of MFCCs are predictably modified by additive noise and the amount of change is related to the noise level. Then, we propose a new supervised method for SNR estimation which is based on a regression model trained on MFCCs of the noisy signals. Experimental results show that the proposed approach provides accurate estimation and consistent performance for various speech types under different noise conditions.

AB - The presence of background noise in signals adversely affects the performance of many speech-based algorithms. Accurate estimation of signal-to-noise-ratio (SNR), as a measure of noise level in a signal, can help in compensating for noise effects. Most existing SNR estimation methods have been developed for normal speech and might not provide accurate estimation for special speech types such as whispered or disordered voices, particularly, when they are corrupted by non-stationary noises. In this paper, we first investigate the impact of stationary and non-stationary noise on the behavior of mel-frequency cepstral coefficients (MFCCs) extracted from normal, whispered and pathological voices. We demonstrate that, regardless of the speech type, the mean and the covariance of MFCCs are predictably modified by additive noise and the amount of change is related to the noise level. Then, we propose a new supervised method for SNR estimation which is based on a regression model trained on MFCCs of the noisy signals. Experimental results show that the proposed approach provides accurate estimation and consistent performance for various speech types under different noise conditions.

KW - Global SNR estimation

KW - MFCC

KW - Pathological voice

KW - Support vector regression

KW - Whispered speech

UR - http://www.scopus.com/inward/record.url?scp=85054227441&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2018.8462459

DO - 10.1109/ICASSP.2018.8462459

M3 - Article in proceeding

SP - 296

EP - 300

BT - 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

PB - IEEE

ER -

Poorjam AH, Little MA, Jensen JR, Christensen MG. A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE. 2018. p. 296-300. 8462459. (I E E E International Conference on Acoustics, Speech and Signal Processing. Proceedings). https://doi.org/10.1109/ICASSP.2018.8462459