Sistema de reconhecimento de locutor integrado a comunicação e controle dos movimentos de um robô humanoide.

Abstract

Speech is one of the most effective and used ways of communication by humans. When adopted in computer systems as a form of input, it makes any human-machine interaction, in principle, more natural. According to this trend, Automatic Speaker Recognition systems were created in order to identify the sender of a particular individual voice signal, by analyzing the characteristics of that signal. One of the applications of these systems, allows us to add differents security levels to communications by authenticating users and their permissions. In view of this, the objective of this work is to develop an automatic speaker recognition into a humanoid robot, which should allow communication only by authorized people. This communication may be done in two ways, one being through commands to control the movement of the robot, and also for the interpretation of a simple question, which the robot must be able to process an answer. In the end, there was obtained an application with acceptable levels of speaker recognition and also it was able to control, through speech, the humanoid movements and providing a simplified communication in form of question and answers. For the extracting features of the speech signal and identifying speaker steps, the methods of Mel-Cepstral Coefficients and Gaussian Mixture Models are applied, which are considered state of art in the literature.