The Use of T-F Masking in Dual-Microphone System

In a variety of speech applications, the use of microphone array for extracting desired signal and suppressing interference noise is common widely. In particular, the effectiveness, caused by exploiting a priori spatial information and adaptive signal processing algorithm, allows saving the useful target signal and attenuating background noise. MVDR is one of the most promising filters that are used in speech equipment. MVDR cancels noise and preserves signal, which come from a certain direction. However, the interference and reverberation are not negligible, the degraded performance is not avoidable. In this paper, the author proposed a combination T-F masking with MVDR filter for increasing the robustness performance in scenario with complex noise. With an estimation of T-F masking, the resulted experiments are validated the suggested method, which is suitable for incorporating into multimicrophone system.