| Abstract [eng] |
Colloquial language are common communication tool for people. Technology is improved and occupied an increasingly prominent place in people's daily activities rapidly, therefore it is very important to improve technologies control by the easiest way for people - by voice. Therefore, the main purpose of automatic speech recognition systems design and development is to develop technologies which could hear, understand, speak and act according to the information received by voice. Technology research is being carried out for years both in Lithuania and abroad. Kaunas University of Technology scientists implemented the project - \"Hybrid recognition technology for voice interface” in 2012-2013. Master thesis is directly associated with researches and results of this project. There was accomplished detailed analysis of the literature covering the colloquial language, systems of voice recognition, speech recognition modelling methods. Researches of speech recognition was accomplished by using other language recognition system and Lithuanian recognition system. There was applied various methods for researches of speech recognition – Hidden Markov Models, Data mining system WEKA, various data mining classificatory and other. This paper presents the \"Voice-Based Codes Recognition system”, which recognizes dictated voice commands. Demonstration program is written in C # programming language. Also there was accomplished voice commands recognition accuracy tests. The obtained results are analysed with conclusions. |
