Uma Melhoria na Convergência do Método ICA-MAP para Remoção de Ruído em Sinal de Voz

Fabio Louvatti do Carmo, Evandro Ottoni Teatini Salles

DOI: 10.14209/sbrt.2013.137

Evento: XXXI Simpósio Brasileiro de Telecomunicações (SBrT2013)

Keywords: Speech Enhancement Maximum a posteriori MAP Independent Component Analysis (ICA) Non-Gaussian Noise Optimization

Abstract

This paper presents a modification to the ICA-MAP algorithm in order to improve its convergence. It was observed through testing that it is possible to limit the magnitude of the gradient vector, used to estimate the denoising model, and thereby improve the stability of algorithm. Such adaptation can be understood as a restriction the original optimization problem.

Download