Speech Enhancement Based on Modulation-Domain Parametric Multichannel Kalman Filtering

Recently we presented a modulation-domain multichannel Kalman filtering (MKF) algorithm for speech enhancement, which jointly exploits the inter-frame temporal evolution of and inter-channel spatial correlation to estimate clean signal. The goal enhancement is suppress noise while keeping undistorted, key problem achieve best trade-off between distortion reduction. In this paper, extend MKF by presenting parametric (PMKF) includes parameter that enables flexible control behaviour in each time-frequency (TF) bin. Based on decomposition cost function, new function PMKF proposed, uses controlling weight reduction terms. An optimal gain derived using minimum mean squared error (MMSE) criterion. We analyse performance proposed MKF, show its relationship weighted Wiener filter (SDW-MWF). To evaluate impact performance, further propose systems adaptively chosen TF Experiments publicly available head-related impulse response (HRIR) database different noisy reverberant conditions demonstrate effectiveness method.