INVESTIGADORES
DI PERSIA Leandro Ezequiel
congresos y reuniones científicas
Título:
Effects of Wiener Filter on Blind Source Separation by processing on sequentially connected bins
Autor/es:
TADASHIGE NOGUCHI; KENKO OHTA; LEANDRO EZEQUIEL DI PERSIA; MASUZO YANAGIDA
Lugar:
Nara
Reunión:
Encuentro; Meeting of the Engeneering Acoustics Group; 2007
Resumen:
Permutation-Free ICA is a method that can avoid so-callled Permutation problem that appears inconventional frequency-domain ICA due to indeterminicity among source identity in each frequency bin.Permutation-Free ICA carries out separation on a set of long vectors consisting of connected temporal changesof frequency components of the received signal. The separation matrix obtained in Permutation-Free ICA,however, has common directivity for all frequency bins. Applying Wiener Filter on a resultant signal of BlindSource Separation, frequency characteristics of the separation matrix are thought to be taken into account. Inthis paper, effects of applying Wiener Filter after Blind Source Separation are evaluated using segmental SNRand PESQ as evaluation indeces.
