Employing stochastic constrained LMS algorithm for ASR frontend processing

In scenarios with multiple input single output systems, the stochastic constrained least mean-squares (LMS) algorithmhas been proven to be an effective approach. However, whenonly two input channels are available, it is unclear whetherthis approach still yields improvements. In this paper, we investigate the stableness and the robustness of the constrained LMS algorithm on "Track 1" of "2nd CHiME Challenge" and show that it leads to small yet consistent improvements on all signal-to-noise settings.