Uso do reconhecimento da fala para aplicação em fechadura eletrônica comandada pela voz

Abstract

The aim of the present study is to use the speech recognition technology to the de velopment of a digit recognition system applied to a prototype of an electronic lock. A theoretical background is described the physiological bases of speech and the ex traction technique of mel-frequency cepstral parameters. The speech recognition system for using in electronic lockers is developed in this study using the speech de coder system Julius in conjunction with the Hidden Markov Model Toolkit, to build and manipulate statistical models called Hidden Markov Models. Next is described the prototype implementation using the Raspberry Pi. The proposed objectives were achieved, and the developed system presented satisfactory operation with different speakers and could be validated in prototype implemented with the Raspberry Pi.