Frequency and Wavelet Filtering for Robust Speech Recognition

Murat Deviren 1 Khalid Daoudi 1
1 PAROLE - Analysis, perception and recognition of speech
INRIA Lorraine, LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
Abstract : Mel-frequency cepstral coefficients (MFCC) are the most widely used features in current speech recognition systems. However, they have a poor physical interpretation and they do not lie in the frequency domain. Frequency filtering (FF) is a technique that has been recently developed to design frequency-localized speech features that perform similar to MFCC in terms of recognition performances. Motivated by our desire to build time-frequency speech models, we wanted to use the FF technique as front-end. However, when evaluating FF on the Aurora-3 database we found some discrepancies in the highly mismatch case. This led us to put FF in another perspective: the wavelet transform. By doing so, we were able to explain the discrepancies and to achieve significant improvements in recognition in the highly mismatch case.
Type de document :
Communication dans un congrès
Kaynak, Okyay and Alpaydin Ethem and Oja, Erkki and Xu, Lei. Artificial Neural Networks and Neural Information Processing - Joint International Conference ICANN/ICONIP2003, 2003, Istanbul, Turquie, Springer-Verlag, 2714, pp.452-462, 2003, Lectures Notes in Computer Science
Liste complète des métadonnées

https://hal.inria.fr/inria-00099753
Contributeur : Publications Loria <>
Soumis le : mardi 26 septembre 2006 - 09:40:57
Dernière modification le : jeudi 11 janvier 2018 - 06:19:57

Identifiants

  • HAL Id : inria-00099753, version 1

Collections

Citation

Murat Deviren, Khalid Daoudi. Frequency and Wavelet Filtering for Robust Speech Recognition. Kaynak, Okyay and Alpaydin Ethem and Oja, Erkki and Xu, Lei. Artificial Neural Networks and Neural Information Processing - Joint International Conference ICANN/ICONIP2003, 2003, Istanbul, Turquie, Springer-Verlag, 2714, pp.452-462, 2003, Lectures Notes in Computer Science. 〈inria-00099753〉

Partager

Métriques

Consultations de la notice

158