Title :
Multiple application of the MLLT based on clustering supported by phonetic knowledge
Author :
J.V. Psutka;Ales Prazak
Author_Institution :
Department of Cybernetics, Faculty of Applied Sciences, University of West Bohemia, Pilsen, Czech Republic
Abstract :
This paper intends to show the effect of multiple use of a Maximum Likelihood Linear Transform (MLLT) in continuous speech recognition tasks. The corresponding covariance matrices that were subjected to the transform were clustered using phonetic knowledge. The impact of this approach was tested on the ASRs with PLP- and MFCC-based front-ends and was compared with other known feature decorrelation and dimension reduction techniques.
Keywords :
"Covariance matrix","Hidden Markov models","Linear discriminant analysis","Decorrelation","Speech recognition","Gaussian distribution","Vectors","Principal component analysis","Cybernetics","Acoustic testing"
Conference_Titel :
Signal Processing, 2008. ICSP 2008. 9th International Conference on
Print_ISBN :
978-1-4244-2178-7
Electronic_ISBN :
2164-523X
DOI :
10.1109/ICOSP.2008.4697207
