ResearchBib Share Your Research, Maximize Your Social Impacts
Sign for Notice Everyday Sign up >> Login

RESEARCH OF MODELS AND METHODS OF SPEECH RECOGNITION IN A WEB SERVICE FOR ORGANIZATION OF USER INFORMATION

Journal: International Scientific Journal "Internauka" (Vol.1, No. 99)

Publication Date:

Authors : ;

Page : 53-57

Keywords : web-service; language recognition systems; client; server; data storage Petri net; graph of achievable markings;

Source : Downloadexternal Find it from : Google Scholarexternal

Abstract

The construction of a web service for organizing user information with the possibility of voice input, speech recognition for more convenient data storage and further use is considered. The study also examines the construction of models and methods of speech recognition systems, the main analogues of the developed web service, their shortcomings and advantages. Among the disadvantages were identified inconvenient and incomprehensible interface, the inability to structure the information into categories, the system of voice input, information retrieval and synthesis into text is not developed. The general structural scheme of the service was built, which shows the connection of client, server and data warehouse, the scheme of algorithm of voice input and synthesis of sounds into text, the scheme of the main stages and methods of speech recognition, such as the method based on the hidden Markovsky models, the method based on dynamic programming, the construction of neural networks and the method based on dictionary search. The latter was taken as a basis for web service development. The web service is divided into three main parts — client, server and data warehouse. The user is also able to store not only text information, but also audio and video files, personal images and pictures. Another advantage of the developed service is the ability to structure information by categories and notebooks, and an intelligent and convenient text editor. There is also the construction of a Petri network and a graph of achievable markings. Petri network shows data exchange between client, server parts and data storage, time of data processing and thanks to statistics shows correct work of the system as a whole. Statistics shows how much data in the form of queries was processed by the system, minimum, maximum, average number of queries, and time of processing them. On its basis, the decision is made to refine the service, i. e. to simplify it in order to speed up its work.

Last modified: 2021-04-20 17:54:37