An Efficient Structured Sparsity Model For Reverberant Speech Separation. int. jour. eng. com. sci 2015, 4 (10).