Design and Evaluation of a Real-Time Speech Recognition System

摘要

Real-time speech recognition systems are in huge demand due to the technological developments in the areas of humanoid robots, voice-based assistive systems, natural language processing, autonomous systems and many more. In this paper, design of a real-time speech recognition system using a novel reduced MFCC feature set and Relevance Vector Machine (RVM) classifier is proposed. The computation time required for each stage of the proposed real-time speech recognition system is analyzed and reported. The performance of proposed system is also evaluated by using Support Vector Machine (SVM) classifier. A maximum recognition accuracy of 99% and 98% is obtained on using SVM and RVM respectively for the proposed system.