IMPROVING DUAL-MICROPHONE SPEECH ENHANCEMENT BY LEARNING CROSS-CHANNEL FEATURES WITH MULTI-HEAD ATTENTION

Hand-crafted spatial features, such as inter-channel intensity difference (IID) and inter-channel phase difference (IPD), play a fundamental role in recent deep learning based dualmicrophone speech enhancement (DMSE) systems. However, learning the mutual relationship between artificially designed spatial feature and spectral feature is hard in the end-to-end DMSE. In this work, a novel architecture for DMSE using multi-head cross-attention based convolutional recurrent network (CRN) is presented. The proposed model includes a channel-independent encoding architecture for spectral estimation and a strategy to extract cross channel features through an multi-head cross attention mechanism. In addition, the proposed approach specifically formulates the decoder with an extra SNR estimator to estimate frame-level SNR under a multi-task learning framework, which is expected to avoid speech distortion lead by end-to-end DMSE module. Finally, a spectral gain function is adopted to further suppress the unnatural residual noise. Experiment results demonstrated a superior performance of the proposed model against several state-of-the-art models.