Ngoc
Duong
INRIA, Centre Inria Rennes - Bretagne Atlantique, France
qduong@irisa.fr


ABOUT THE ALGORITHM:

- The vector STFT coefficient of the spatial images of each source is modeled as a zero-mean Gaussian variable with full-rank time- and
frequency-dependent covariance [1].

- Two methods have been investigated for improving the accuracy of parameter estimation, based on the
use of local observed covariance and auditory motivated time-frequency representation [2].

- An iterative expectation-maximization (EM) algorithm with a suitable initialization scheme is derived to
estimate model parameters [2].


REFERENCES:

1. N. Q. K. Duong, E. Vincent, and R. Gribonval, "Under-determined reverberant audio source separation using a
full-rank spatial covariance model", IEEE Transactions on Audio, Speech and Language Processing, 
Special Issue on Processing Reverberant Speech. To appear 2010.
 
2. N. Q. K. Duong, E. Vincent, and R. Gribonval, "Under-determined reverberant audio source separation using local
observed covariance and auditory-motivated time-frequency representation", Proc. ICA/LVA 2010, submitted