Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Applied Methods for Blind Speech Enhancement
Responsible organisation
2008 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Acoustic disturbances influence human speech communication by interfering with the communication process. In the worst case, it is impossible to communicate at all due to these disturbances. Methods that reduce the influence of the disturbances while preserving speech intelligibility are often desired. This thesis proposes real-world solutions for applied speech enhancement using autonomous and robust methods. Most of the work of the thesis concerns solutions to the problem of reducing acoustic disturbances within the framework of Blind Speech Enhancement (BSE). Notably, the term "blind" is assigned a positive attribute as it implies that the speech enhancement is carried out without any explicit references required. Instead, an assumption about the statistical independence between the sources coupled with an assumption regarding distinguishing statistical properties of the sources underpin the proposed methods. The unifying theory is Independent Component Analysis (ICA), which is performed by means of spatial filtering. Two of the methods that are proposed in this thesis are shown, both in a theoretical and an empirical framework, to be robust in a real application while preserving stability even for Gaussianonly sources. Existing methods cannot guarantee stability in this scenario and Gaussian-only source mixtures may be the case in a real environment. The difference between the two methods lies in the different optimization strategies and the introduced approximations. The idea of injecting a single-channel method into the control loop of a blind beamformer is also proposed. In particular, two approaches are derived that aim at improving the blind beamformer in the case of disturbing noise and maintaining the same performance for different signal input levels. Finally, implementation aspects of a single-channel speech enhancer are discussed. The implementation aspects deal with the implementation of a speech enhancer in several different platforms such as analogue hardware, digital hardware, as well as hybrid analogue and digital hardware.

Place, publisher, year, edition, pages
Karlskrona: Blekinge Institute of Technology , 2008. , p. 231
Series
Blekinge Institute of Technology Doctoral Dissertation Series, ISSN 1653-2090 ; 15
National Category
Signal Processing
Identifiers
URN: urn:nbn:se:bth-00434Local ID: oai:bth.se:forskinfo956124B733F4774BC125754600358B10ISBN: 978-91-7295-154-9 (print)OAI: oai:DiVA.org:bth-00434DiVA, id: diva2:835948
Available from: 2012-09-18 Created: 2009-01-22 Last updated: 2015-06-30Bibliographically approved

Open Access in DiVA

fulltext(1616 kB)282 downloads
File information
File name FULLTEXT01.pdfFile size 1616 kBChecksum SHA-512
450ce4d2939bc2c0b57b4eb32d7d3aaa3bb5bc97c8bf53b89f7a0d0b3e3c40d0a33c5fda76d7deedc51001ce09381c153119020d10cffc51752ab06b296c1344
Type fulltextMimetype application/pdf

Signal Processing

Search outside of DiVA

GoogleGoogle Scholar
Total: 282 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 210 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf