بهبود جداسازي سيگنال منابع به روش كور با استفاده از زاويه منابع

IMPROVEMENT OF BLIND SOURCE SEPARATION USING SOURCES ANGLES

In this paper we present a method for improving Blind Source Separation and decreasing of Musical noise. In proposed method we separate the signals in multiple stage with using ICA algorithm and continuous masks which are calculated by sources angles. In simulation section with different examples we demonstrate declining of noise and enhancement of quality of separated signals. In this paper the Hard_DESPRIT is used as a reference.