Open AccessOn the Possibility of Embedding the Mechanism of Linguistic Anticipation into Speech Recognition Systems
Author(s) -
Daniel Kurushin,
N. M. Nesterova,
Olga Soboleva
Publication year - 2021
Publication title -
journal of information technology and applications
Language(s) - English
Resource type - Journals
eISSN - 2233-0194
pISSN - 2232-9625
DOI - 10.7251/jit2102099k
Subject(s) - anticipation (artificial intelligence) , computer science , quality (philosophy) , speech recognition , mechanism (biology) , reading (process) , similarity (geometry) , artificial intelligence , natural language processing , linguistics , image (mathematics) , philosophy , epistemology
The paper deals with the problems of modeling speech recognition systems. The authors proposed to use the mechanism of linguistic anticipation in the speech recognition systems. It is known that anticipation is a kind of phenomenon of anticipatory reflection, which can provide an opportunity for the subject to “look into the future.” Anticipation is believed to be an effective method of improving reading technique in children as it enables to increase the speed of reading [1]. The similarity of the learning processes of the human brain and artificial neural-like algorithms allows to suggest that the inclusion of anticipation mechanisms into the operation of the speech recognition algorithm can improve the quality of the system. The paper presents the experiment carried out with the purpose to study the probability of increasing the quality of modern speech recognition systems provided that linguistic anticipation is embedded into such a system. The obtained results are discussed and possible directions for further work on this topic are considered.