Separação cega de fontes: aplicando um algoritmo para decomposição em vectores próprios

Abstract

This work presents an iterative algorithm for blind source separation. The algorithm is based on the generalized eigendecomposition of a matrix pencil. A Matlab package of software was developed to evaluate the algorithm performance. Some experimental results, using artificially mixed audio signals and a matrix pencil computed at the input and at the output of a linear finite impulse response, are also presented.