Frequency‐domain integration of acoustic echo cancellation and a generalized sidelobe canceller with improved robustness

Abstract

Abstract. For hands-free acoustic humadmachine interfaces, as required, e.8.. for automatic speech recognition, teleconferencing, and other multimedia services. microphone arrays using generalized sidelobe cancellers (GSCs) in conjunction with acoustic echo cancellation (AEC) can be efficiently applied for optimum communication. This contribution first devises a new structure for combining AEC and GSC in the frequency domain. We show that computational complexity is reduced by more than a factor of ten compared to a time-domain mngement. Second, robustness issues of the GSC adaptation mechanism are addressed. We illustrate how robust adaptation is assured in the new structure. Third, we propose alternatives for the fixed beamformer part of the GSC, which allow to specify (a) the width of the target tracking region and (b) improve the suppression of low frequencies. GSC robustness constraints are included using linear optimization with configurable penalty functions.

Analysis of dual-channel ICA-based blocking matrix for improved noise estimation

“…For this simulation, an ideal case is assumed, i.e., the microphone signal contains only the target signal. The simulated ABM can be regarded as an ideal version (in a supervised case) of a conventional ABM using an LMS algorithm as proposed in [23,24].…”

“…However, the conventional LCMV/MVDR BM will likely lead to target signal leakage as it is conceived for time-invariant scenarios, and any movement of the target source will lead to a steering error relative to the true DoA for the target signal and its reflections. To improve the robustness against the steering error, an adaptive BM was proposed in [23,24], which needs an adaptive control requiring source activity information. In [25], the relative transfer function (RTF)-based BM for LCMV/MVDR beamforming was proposed.…”

Adaptive Beamforming for Audio Signal Acquisition

“…The GSC has been applied to audio signal processing, e.g., in [1], [8], [64], [65], [66], [67], [68], [69].…”

Adaptive Beamforming with a Maximum Negentropy Criterion

“…Therefore, Hoshuyama's algorithm requires in some sense a trade-off between the avoidance of signal cancellation and suppression of the interference signals. This problem can be solved by simply halting the adaptation of the ABM and only updating the active weight vectors in the case of a high signal-to-noise ratio (SNR) [13]. Such a switching algorithm is based on SNR, however, and requires complicated rules which must generally be determined empirically.…”

“…• updating the active weight vector only when noise signals are dominant [6], [7], [8]; • constraining the update formula for the active weight vector with the leaky least mean square (LMS) algorithm [9], [10] or with power of outputs of the blocking matrix [11]; • using multi-channel target signals received by the microphone array and correlation matrices of the clean and noise corrupted target signals in a calibration phase, [12]; • blocking the leakage of desired signal components into the sidelobe canceller by appropriately designing the blocking matrix [11], [13], [14], [15]; • taking speech distortion due to the leakage of a target signal into account using a multi-channel Wiener filter which aims at minimizing a weighted sum of residual noise and speech distortion terms [16]; and • using acoustic transfer functions from a desired source to microphones instead of merely compensating for the time delays [8], [15], [17], [18].…”