Uzbek speech commands recognition and implementation based on HMM

Abstract

In developed countries, many effective speech recognition systems have been developed using advanced technologies. But this speech recognition systems are not suitable for Uzbek language or the level of recognition of speech commands may be low. To solve this problem, the article discusses the stages of recognition of speech commands and intelligent processing algorithms. Based on the hidden Markov model, a recognition system for Uzbek speech commands was implemented in Python using CMU Sphinx technology that recognizes Uzbek speech commands on a computer and Raspberry Pi3 devices. The accuracy of the recognition system for Uzbek speech commands are showed 80-95%.