A block least squares approach to acoustic echo cancellation

Abstract

We propose an efficient block least-squares (BLS) algorithm for acoustic echo cancellation. The high computation and memory requirements associated with a long room echo make the simple, gradient-based LMS filter a more acceptable commercial solution than a full-fledged LS canceler. However, the LMS echo canceler has slower convergence and worse steady-state performance than its LS counterpart. In the proposed BLS approach, the autocorrelation and cross-correlation of the source and echo, required in solving the LS normal equations, are performed once per block using W s . With appropriate data windowing the autocorrelation matrix is constrained to be Toeplitz, allowing the corresponding normal equations to be solved efficiently. The positive definiteness of the autocorrelation function eliminates the stability problems of other fast LS algorithms. BLS can reduce the echo residual to the level of background noise, allowing a residual power based, statistical near-end speech detector to be devised. Performance in real environments under various settings of filter length, SNR, near-end speech presence, etc., is investigated.

Multispeaker speech activity detection for the ICSI meeting recorder

“…Crosscorrelation analysis suggests a different approach to the problem of crosstalk, namely, estimating the coupling between different channels and using the estimates to cancel the crosstalk signals. We are investigating such an approach based on the Block Least Squares algorithm described in [7]. However, the situation is complicated by the very rapid changes in coupling that occur when speakers or listeners move their heads.…”

Spectral Domain B-spline Identification in Acoustic Echo Cancellation

“…If basis functions are complex harmonics, the echo path is modeled as a FIR filter with Nϕ filter taps, R is the auto-correlation matrix of the excitation signal, and ξ is the cross-correlation vector of the excitation and microphone signals. This case corresponds to the block LS approach adopted in [3]. Unfortunately, for harmonic basis functions, the matrix R in general is not sparse, which makes solving the system (7) with large Nϕ a complicated problem.…”

“…The vectors α = [α(1), α(2), α (3)] and ρ = [ρ(1), ρ(2), ρ (3)] are chosen to obtain the best cancellation performance. Other dependencies αt on ε 2 and Ey can also be used.…”

“…The Fast AP algorithm has been proposed [2], but it demonstrates numerical instability and it is sensitive to noise. Good cancellation performance is achieved by using the least squares (LS) block approach [3]. However, even with the Toeplitz approximation of normal equations in the LS problem and use of the computationally efficient Levinson algorithm, the complexity of the echo canceler is still high.…”