“…Further, if any of the signals x(k) and v(k) are white, then ρ(k) = 0. In the case of speech, the magnitude of the first factor in equation (14) is likely to decrease as i increases, since the autocorrelation of a speech signal usually decrease rapidly as the lag increases [12], [13], [14]. Further, since all eigenvalues if X i are non-negative and ≤ 1, the magnitude of the eigenvalues of the matrix resulting of the product i−1 j=0 X j are monotonically decreasing as i increases.…”