Presentation 2005/2/25
A Neural Network Model of Speech Recognition Based on The Gestalt Principle
Hironobu OASHI, Akinao YOSHIDA, Yoko YAMAGUCHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The gestalt principle has been known to be applicable to auditory recognitions. We hypothesize that phonological perception in speech recognition is a process to organize a good form of segments (morae in Japanese) following to the gestalt principle. Feedforward and backward interactions between a noise and a formant might complete clear and collective formant transitions in morae. By using simulation of a neural network model with noise and formant dependent interactions, we found that a mora with a clear formant transitions can be generated by self-organization of neural activities.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Gestalt / Speech recognition / Segmentalization / Interaction between the features / Backward interaction
Paper # SP2004-165
Date of Issue

Conference Information
Committee SP
Conference Date 2005/2/25(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Speech (SP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Neural Network Model of Speech Recognition Based on The Gestalt Principle
Sub Title (in English)
Keyword(1) Gestalt
Keyword(2) Speech recognition
Keyword(3) Segmentalization
Keyword(4) Interaction between the features
Keyword(5) Backward interaction
1st Author's Name Hironobu OASHI
1st Author's Affiliation Graduate School of Science and Engineering, Tokyo Denki University()
2nd Author's Name Akinao YOSHIDA
2nd Author's Affiliation Graduate School of Science and Engineering, Tokyo Denki University
3rd Author's Name Yoko YAMAGUCHI
3rd Author's Affiliation Graduate School of Science and Engineering, Tokyo Denki University:RIKEN Brain Science Institute:CREST JST
Date 2005/2/25
Paper # SP2004-165
Volume (vol) vol.104
Number (no) 696
Page pp.pp.-
#Pages 6
Date of Issue