An Optimal Speech Recognition Module for Patient's Voice Monitoring System in Smart Healthcare Applications

During recent years, health care domain has rapidly developed in which patients and medical resources are directly connected with the smart way that enables Smart Health Care. The growth in design and development of a speech automated system will provide a life assistant service in smart health care environment. In automating the speech system, speech recognition is one of the basic steps to understand the human recognition and their behaviors. These speech recognition systems will be very much accessible for speakers who suffer from dysarthria, a neurological disability that damages the control of motor speech articulators. In this paper, the main objective is to develop an efficient speech recognition module based on the Voice Input Voice Output Communication Aid (VIVOCA) architecture that can device a support aid to the people with DYSARTHRIA. Totally there are seven features extracted from each noise eliminated real time bilingual isolated word speech signal data uttered by a speaker both in Tamil and English languages. Vector Quantization based Genetic Algorithm codebook is created for the recognition modeling. Optimization of Hidden Markov Model (HMM) is done based on Particle Swarm Optimization (PSO) method to improve the recognition accuracy compared to the conventional HMM and also experiment results of the proposed module shows 95% of accuracy. The proposed module will be very much useful for developing a speech recognition system that facilitates the patients and persons with special needs for communication. The proposed module is also evaluated for its complexity which will be therefore efficient for low consumption of energy.