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...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Russian electrical engineering Jg. 91; H. 11; S. 669 - 672
Hauptverfasser: Bobkov, S. A., Kurushin, D. S., Perevalov, A. M., Soboleva, O. V., Faizrakhmanov, F. A.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Moscow Pleiades Publishing 01.11.2020
Springer Nature B.V
Schlagworte:
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-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung: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.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1068-3712
1934-8010
DOI:10.3103/S1068371220110036