Using Linguistic Anticipation to Improve the Quality of Speech Recognition in Robotic Systems

Anticipation is a phenomenon of a kind of forward-looking reflection that sometimes allows the subject to “see” the future. A number of experts believe that anticipation is an effective method for improving reading skills in children. Effective use of anticipation can significantly increase a person...

Full description

Saved in:
Bibliographic Details
Published in:Russian electrical engineering Vol. 91; no. 11; pp. 669 - 672
Main Authors: Bobkov, S. A., Kurushin, D. S., Perevalov, A. M., Soboleva, O. V., Faizrakhmanov, F. A.
Format: Journal Article
Language:English
Published: Moscow Pleiades Publishing 01.11.2020
Springer Nature B.V
Subjects:
Algorithms
Engineering
Machine learning
Machines
Manufacturing
Processes
Speech recognition
Voice recognition
algorithms for language model usage
speech recognition
natural language processing
anticipation
ISSN:1068-3712, 1934-8010
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Anticipation is a phenomenon of a kind of forward-looking reflection that sometimes allows the subject to “see” the future. A number of experts believe that anticipation is an effective method for improving reading skills in children. Effective use of anticipation can significantly increase a person’s reading speed. The similarity of human learning processes and artificial neural algorithms suggests that introducing “anticipation”-based mechanisms into speech recognition algorithms can improve the quality of these systems. The article discusses the development of algorithms for modeling anticipation in speech recognition. It describes the proposed model and the setup of an experiment aimed at assessing the effect of anticipation mechanisms on the recognition result and discusses experimental findings and possible areas for further research on the topic.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1068-3712
1934-8010
DOI:10.3103/S1068371220110036