Estudio del procesamiento de comandos de voz empleando los modelos ocultos de Markov, aplicado al control autónomo de una silla de ruedas.

Abstract

In this project an investigation is carried out on systems for the recognition of voice commands, which is adapted to a wheelchair, this approach is carried out taking into account the population with motor disability, which is affected by the Exclusive society in terms of providing services, spaces, work, etc. The purpose of our work lies in the development of an algorithm that summarizes the optimal methodologies, extracting the necessary techniques for a good processing of acoustic signals, capable of filtering inaudible and harmful frequency signals for the human being, where the range of usefulness is from 20Hz to 20kHz, this algorithm is capable of recognizing no less than 12 fundamental characteristics of the voice spectrum, in order to identify the phonemes of the words. Among the methods that are part of the system is the Hidden Markov Models, which are statistical comparisons.