Blind source separation for convolutive mixtures
Signal Processing
Neural Networks: A Comprehensive Foundation
Neural Networks: A Comprehensive Foundation
IWANN '01 Proceedings of the 6th International Work-Conference on Artificial and Natural Neural Networks: Bio-inspired Applications of Connectionism-Part II
Hi-index | 0.00 |
In this paper we propose a new strategy to separate convolutive mixtures of temporally-white signals. The basic idea is to transform the convolutive mixture in several instantaneous mixtures by using the discrete Fourier transform. Subsequently, each instantaneous mixture is separated using a neural network whose coefficients are adapted by minimizing the mean squared error between its outputs and a desired signal previously obtained using an unsupervised algorithm (like JADE). This new strategy does not suffer from the amplitude/permutation indeterminacies that appear in other frequency-domain approaches.