Skip navigation
Skip navigation

Single-channel speech dereverberation in noisy environment for non-orthogonal signals

Fahim, Abdullah; Samarasinghe, Prasanga; Abhayapala, Thushara

Description

The detrimental effect of speech reverberation reduces speech quality, limits the performance of automatic speech recognition systems and impairs hearing aids. Spectral enhancement (SE) is a popular method for suppressing the late reverberation and background noise. However, conventional SE-based approaches assume orthogonality between the desired and undesired signal components. This orthogonality assumption does not hold true in most of the practical cases due to a limited time-domain support...[Show more]

dc.contributor.authorFahim, Abdullah
dc.contributor.authorSamarasinghe, Prasanga
dc.contributor.authorAbhayapala, Thushara
dc.date.accessioned2020-07-10T00:24:20Z
dc.date.available2020-07-10T00:24:20Z
dc.identifier.issn1610-1928
dc.identifier.urihttp://hdl.handle.net/1885/205999
dc.description.abstractThe detrimental effect of speech reverberation reduces speech quality, limits the performance of automatic speech recognition systems and impairs hearing aids. Spectral enhancement (SE) is a popular method for suppressing the late reverberation and background noise. However, conventional SE-based approaches assume orthogonality between the desired and undesired signal components. This orthogonality assumption does not hold true in most of the practical cases due to a limited time-domain support and the short-time stationarity of the speech signals, and thereby, affects estimation accuracy. To circumvent this issue, Lu et al. relaxed the orthogonality assumption by proposing a geometric approach to spectral subtraction (GSS) and evaluated their algorithm against different kinds of background noise. In our work, we comprehensively analyze the model by virtue of a simplified GSS transfer function to gain an insight into the algorithm. We conduct a series of experiments to validate GSS and explore its limitations in diverse realistic scenarios with both reverberation and background noise through a comprehensive end-to-end system for speech dereverberation and noise suppression. We also analyze the performance of GSS using the experimental data of the 2014 REVERB challenge and compare it with other conventional approaches such as spectral subtraction, Wiener Filter, minimum mean square error short-time spectral amplitude estimator and log spectral amplitude estimator, as well as with the contemporary methods of the 2014 REVERB challenge.
dc.description.sponsorshipAustralian Research Council: DP140103412
dc.format.mimetypeapplication/pdf
dc.language.isoen_AU
dc.publisherS. Hirzel Verlag
dc.rights© 2018 S. Hirzel Verlag
dc.sourceActa Acustica united with Acustica
dc.titleSingle-channel speech dereverberation in noisy environment for non-orthogonal signals
dc.typeJournal article
local.description.notesImported from ARIES
local.identifier.citationvolume104
dc.date.issued2018-11-01
local.identifier.absfor090609 - Signal Processing
local.identifier.ariespublicationu4485658xPUB2658
local.publisher.urlhttps://www.ingentaconnect.com/
local.type.statusAccepted Version
local.contributor.affiliationFahim, Abdullah, College of Engineering and Computer Science, ANU
local.contributor.affiliationSamarasinghe, Prasanga, College of Engineering and Computer Science, ANU
local.contributor.affiliationAbhayapala, Pallage (Thushara), College of Engineering and Computer Science, ANU
dc.relationhttp://purl.org/au-research/grants/arc/DP140103412
local.bibliographicCitation.issue6
local.bibliographicCitation.startpage1041
local.bibliographicCitation.lastpage1055
local.identifier.doi10.3813/AAA.919270
local.identifier.absseo970109 - Expanding Knowledge in Engineering
dc.date.updated2020-06-23T00:56:06Z
local.identifier.scopusID2-s2.0-85056764189
dcterms.accessRightsOpen Access
dc.provenancehttp://v2.sherpa.ac.uk/id/publication/23848..."Author Accepted Manuscript can be made open access on insitutional repository after 12 month embargo" from SHERPA/RoMEO site (as at 10/7/20).
CollectionsANU Research Publications

Download

File Description SizeFormat Image
01_Fahim_Single-channel_speech_2018.pdf1.09 MBAdobe PDFThumbnail


Items in Open Research are protected by copyright, with all rights reserved, unless otherwise indicated.

Updated:  17 November 2022/ Responsible Officer:  University Librarian/ Page Contact:  Library Systems & Web Coordinator