Diseño de un sistema de reconocimiento de voz basado en la red neuronal de KOHONEN y software libre, para la gestión del Sistema domótico HDL BUS PRO.

Abstract

The present work research has the aim to develop a voice recognition system in general, which captures the words spoken by the user through a microphone to process them and compare with existing words in a previously recorded database. The purpose of this project was to control the domotic system HDL-BUS PRO through voice commands, allowing people with motor disabilities to achieve greater autonomy. The voice signals were processed by digital filters and extracted their features with the help of the discrete transform of Wavelet, the obtain characteristics was entered in a neuronal network based on the model KOHONEN to identify the spoken word by the user. The results obtained of the experiments shown that the system of voice recognition developing fulfil with the proposal, it was recommended to increase the database of the training patterns of neuronal network to have a more precision in the voice recognition system.