Sistema de automação residencial para deficientes visuais baseado em reconhecimento de voz

Abstract

The present text aims to develop a prototype of home automation system for blind users based on voice recognition using artificial neural networks embedded on mobile devices such as mobile phones and tablets for example, simple, easy to use and based on free and low cost platform such as MIT AppInventor, Google Cloud API and Bluetooth communication with Arduino device. The system presents a good hit rate of commands recognition with different users with different voice tones and under different scenarios, such as different distances and quiet and noisy ambients, although not having great results in scenarios on which the device was far from the user and the ambient was noisy, proving itself as a viable and easy to implement and to replicate solution. In addition, with the intention of ensuring the user the command has been given, the commands are confirmed via sound signals.