Utilization of Multiple Units in Human And Machine Recognition of Continuous Speech---- Perceptual Evidence And A Proposal For An Asr System

Auteurs-es

  • H. Fujisaki
  • H. Udagawa
  • N. Kanedera

Résumé

The ultimate goal of automatic speech recognition (ASR) is obviously to replicate the human capability of speech processing by machine. Research of ASR will thus profit very much from investigations into the human processes of speech perception/comprehension. Few studies, however, seem to have been made along this line. The present paper summarizes a series of psycholinguistic experiments conducted to elucidate certain aspects of human speech perception, especially in relation to the units of processing. On the bases of these experiments, we propose a new system for continuous speech recognition utilizing multiple units.

Fichiers supplémentaires

Publié-e

2022-12-03

Comment citer

1.
Fujisaki H, Udagawa H, Kanedera N. Utilization of Multiple Units in Human And Machine Recognition of Continuous Speech---- Perceptual Evidence And A Proposal For An Asr System. Canadian Acoustics [Internet]. 3 déc. 2022 [cité 4 déc. 2024];14(3 bis):28-9. Disponible à: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/3507

Numéro

Rubrique

Actes du congrès de la Semaine canadienne d'acoustique