Concerning the parameters choice of a speech recognition system of isolated Arabic words

Author

Senani, CH ; Bedda, M. ; Bourouba, H. ; Corba, Med Amara

Author_Institution

Dept. of electron., Univ. Badji-Mokhtar, Annaba, Algeria

fYear

2004

fDate

19-23 April 2004

Firstpage

471

Lastpage

472

Abstract

In this paper basic technologies intervening in the design of recognition system of Arab isolated words based on statistical modeling by the continuous HMMs (hidden Markov model) is presented. A comparative study based on the methods of analysis for the parameterization phase and the effect of the number of Gaussians and states for the training phase is studied. The best rate of recognition is obtained by the use of the differential parameters of the first order (MFCC).

Keywords

Gaussian processes; hidden Markov models; speech recognition; statistical analysis; Gaussian states; continuous HMM; first order differential parameters; hidden Markov model; isolated Arabic words; parameterization phase; speech recognition system; statistical modeling; training phase; Hidden Markov models; Isolation technology; Iterative algorithms; Linear predictive coding; Mel frequency cepstral coefficient; Probability; Signal processing; Space technology; Speech processing; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Information and Communication Technologies: From Theory to Applications, 2004. Proceedings. 2004 International Conference on

Print_ISBN

0-7803-8482-2

Type

conf

DOI

10.1109/ICTTA.2004.1307836

Filename

1307836