Bayesian noise estimation in the modulation domain
Bayesian noise estimation in the modulation domain
Modulation domain has been reported to be a better alternative to time-frequency domain for speech enhancement, as speech intelligibility is closely linked with the modulation spectrum. Motivated by that, this paper investigates the use of modulation domain to model the noise density function. Results show that the modulation domain based Gamma density function better represents the noise density for all time-varying noise signals compared to the non-modulation domain. The modulation based Gamma density is then used to derive noise estimator via a Bayesian motivated MMSE approach. As the Gamma density closely matches the true noise spectrum in the modulation domain, the proposed noise estimator does not require bias compensation even for poor signal-to-noise ratio (SNR) conditions, i.e., ≤ 5 dB. The proposed method yields better noise suppression compared to the state of the art methods and provides higher improvements.
Analysis-modification-synthesis (AMS) framework, Gamma distribution, Minimum mean-square-error (MMSE), Modulation domain, Noise estimation, Single channel speech enhancement
81-92
Singh, Maneesh K.
99c21be7-e456-45dc-a71a-e92828959486
Low, Siow Yong
8fd903a4-b0b0-4c1c-9cc7-c2fe87109376
Nordholm, S.
d2441721-2cf0-4387-a95d-7cd2b956c014
Zang, Zhuquan
6b5a21fc-7a64-4b28-a9b3-2517044e8056
1 February 2018
Singh, Maneesh K.
99c21be7-e456-45dc-a71a-e92828959486
Low, Siow Yong
8fd903a4-b0b0-4c1c-9cc7-c2fe87109376
Nordholm, S.
d2441721-2cf0-4387-a95d-7cd2b956c014
Zang, Zhuquan
6b5a21fc-7a64-4b28-a9b3-2517044e8056
Singh, Maneesh K., Low, Siow Yong, Nordholm, S. and Zang, Zhuquan
(2018)
Bayesian noise estimation in the modulation domain.
Speech Communication, 96, .
(doi:10.1016/j.specom.2017.11.008).
Abstract
Modulation domain has been reported to be a better alternative to time-frequency domain for speech enhancement, as speech intelligibility is closely linked with the modulation spectrum. Motivated by that, this paper investigates the use of modulation domain to model the noise density function. Results show that the modulation domain based Gamma density function better represents the noise density for all time-varying noise signals compared to the non-modulation domain. The modulation based Gamma density is then used to derive noise estimator via a Bayesian motivated MMSE approach. As the Gamma density closely matches the true noise spectrum in the modulation domain, the proposed noise estimator does not require bias compensation even for poor signal-to-noise ratio (SNR) conditions, i.e., ≤ 5 dB. The proposed method yields better noise suppression compared to the state of the art methods and provides higher improvements.
This record has no associated files available for download.
More information
Accepted/In Press date: 19 November 2017
e-pub ahead of print date: 21 November 2017
Published date: 1 February 2018
Keywords:
Analysis-modification-synthesis (AMS) framework, Gamma distribution, Minimum mean-square-error (MMSE), Modulation domain, Noise estimation, Single channel speech enhancement
Identifiers
Local EPrints ID: 417448
URI: http://eprints.soton.ac.uk/id/eprint/417448
ISSN: 0167-6393
PURE UUID: 57e92d66-7fa4-4ec4-9f6d-02608f81d663
Catalogue record
Date deposited: 31 Jan 2018 17:30
Last modified: 15 Mar 2024 18:10
Export record
Altmetrics
Contributors
Author:
Maneesh K. Singh
Author:
Siow Yong Low
Author:
S. Nordholm
Author:
Zhuquan Zang
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics