ivabss.m

Natural Gradient algorithm for Frecuency Domain Blind source separation based on Independent Vector Analysis

[y, W] = ivabss(x, nfft, maxiter, tol, eta, nsou)
 y : separated signals (nsou x N)
 W : unmixing matrices (nsou x nmic x nfft/2+1)
 x : observation signals (nmic x N),
       where nsou is # of sources, nmic is # of mics, and N is # of time frames
 nfft : # of fft points (default =1024)
 eta : learning rate (default =0.1)
 maxiter : # of iterations (default =1000)
 tol : When the difference of objective is less than tol,
           the algorithm terminates (default =1e-6)
 nsou : # of sources (default =nmic)

fiva.m

Fast algorithm for Frecuency Domain Blind source separation based on Independent Vector Analysis

[y, W] = fivabss(x, nfft, maxiter, tol, nsou)
 y : separated signals (nsou x N)
 W : unmixing matrices (nsou x nmic x nfft/2+1)
 x : observation signals (nmic x N),
       where nsou is # of sources, nmic is # of mics, and N is # of time frames
 nfft : # of fft points (default =1024)
 maxiter : # of iterations (default =1000)
 tol : When the increment of likelihood is less than tol,
           the algorithm terminates (deault =1e-6)
 nsou : # of sources (default =nmic)

TO-DO

[1] Taesu Kim, "Independent Vector Analysis" Ph.D. Dissertation, KAIST, 2007

[2] Taesu Kim, Hagai Attias, Soo-Young Lee, Te-Won Lee, "Blind source separation exploiting higher-order frequency dependencies" IEEE Transactions on Audio, Speech, and Language Processing 15 (1), 2007

[3] Intae Lee, Taesu Kim, Te-Won Lee, "Fast fixed-point independent vector analysis algorithms for convolutive blind source separation" Signal Processing 87 (8), 2007

[4] Taesu Kim, Torbjørn Eltoft, Te-Won Lee, "Independent vector analysis: An extension of ICA to multivariate components" International Conference on Independent Component Analysis and Signal Separation, 2006

[5] Taesu Kim, Intae Lee, Te-Won Lee, "Independent vector analysis: definition and algorithms", Fortieth Asilomar Conference on Signals, Systems and Computers, 2006