Speech recognition via STT API for autonomous mobile robot

Abstract

This paper deals with speech recognition system based on STT API (Speech To Text Application Programming Interface), which was mainly designed for mobile robots Human-Machine interface. This paper is based on preliminary work where one word commands were used. The solution presented in this paper uses the pre-learned database with Bayesian approach. This solution increases the success rate in command recognition against to use plain STT API itself. The major extension of previous work is recognition of whole sentences instead of one single word command. Experimental results with complete sentences are presented in this paper.