A neural network approach to the dimensionality of the perceptual vowel space

Auteurs-es

  • Terrance M. Nearey Dept. of Linguistics, University of Alberta, Edmonton, Alta. T6G 2E7, Canada
  • Michael Kiefte School of Human Comm. Disorders, Dalhousie University, Halifax, NS B3H 1R2, Canada

Mots-clés :

Computer simulation, Linguistics, Mathematical models, Neural networks, Speech analysis, Speech coding, Transfer functions, Two dimensional, Dummy-variable coding, Monophthongs, Perceptual vowel space

Résumé

A novel method to examine the degree to which a two- or three-dimensional space can accurately represent listeners' perception of a large three-formant vowel continuum was examined. As such, the results showed that no two-dimensional representation adequately accounts for listeners' behavior. The key to the analysis was a neural network architecture that will be capable of implementing an optimal, non-parametric, two-dimensional representation of the stimulus space.

Fichiers supplémentaires

Publié-e

2003-09-01

Comment citer

1.
Nearey TM, Kiefte M. A neural network approach to the dimensionality of the perceptual vowel space. Canadian Acoustics [Internet]. 1 sept. 2003 [cité 15 févr. 2025];31(3):16-7. Disponible à: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/1527

Numéro

Rubrique

Actes du congrès de la Semaine canadienne d'acoustique

Articles les plus lus du,de la,des même-s auteur-e-s

1 2 > >>