Data driven method for non-intrusive speech intelligibility estimation

Author

Sharma, Dushyant ; Hilkhuysen, Gaston ; Gaubitch, Nikolay D. ; Naylor, Patrick A. ; Brookes, Mike ; Huckvale, Mark

Author_Institution

Centre for Law Enforcement Audio Res. (CLEAR), Imperial Coll. London, London, UK

fYear

2010

fDate

23-27 Aug. 2010

Firstpage

1899

Lastpage

1903

Abstract

We propose a data driven, non-intrusive method for speech intelligibility estimation. We begin with a large set of speech signal specific features and use a dimensionality reduction approach based on correlation and principal component analysis to find the most relevant features for intelligibility prediction. These are then used to train a Gaussian mixture model from which the intelligibility of unseen data is inferred. Experimental results show that our method gives a correlation with subjective intelligibility of 0.92 and a correlation of 0.96 with the ANSI standard Speech Intelligibility Index.

Keywords

Gaussian processes; correlation methods; mixture models; principal component analysis; speech intelligibility; ANSI standard speech intelligibility index; Gaussian mixture model; correlation analysis; data driven method; dimensionality reduction approach; nonintrusive speech intelligibility estimation; principal component analysis; speech signal specific features; subjective intelligibility; Correlation; Estimation; Feature extraction; Noise measurement; Speech; Speech processing; Training;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing Conference, 2010 18th European

Conference_Location

Aalborg

ISSN

2219-5491

Type

conf

Filename

7096447