Regresar

Sound source identification by neural networks

Abstract:

Sound source identification is an intrinsic process that helps people to interact with the environment around them, allowing to interpret the information transmitted by humans, animals and machines, for this reason, it is mimicked by different types of autonomous machines. This article presents the development and implementation of a new method to identify the origin of the sound source for a small robotic system based on three microphones, using two types of artificial neural network, the first one-ADALINE structure, used to obtain a noise environmental filter and the second one-BACKPROPAGATION's structure to identify the source sound in a range between 0 and 180 degrees employing. The identification based on differents tests permit us to have an error of 7% in the last stage. The present project was developed with NI MyRIO device and the software LabVIEW and Matlab.