WPD++: An Improved Neural Beamformer for Simultaneous Speech Separation and Dereverberation
This work provides an incremental improvement in speech enhancement for ASR systems, benefiting users in noisy multi-talker environments.
This paper addresses the problem of simultaneously separating speech from multiple interfering speakers, additive noise, and reverberation to improve automatic speech recognition (ASR). The proposed WPD++ neural beamformer, an improvement over the WPD beamformer, achieves superior enhanced speech quality and lower word error rates (WER) for ASR compared to existing state-of-the-art methods.
This paper aims at eliminating the interfering speakers' speech, additive noise, and reverberation from the noisy multi-talker speech mixture that benefits automatic speech recognition (ASR) backend. While the recently proposed Weighted Power minimization Distortionless response (WPD) beamformer can perform separation and dereverberation simultaneously, the noise cancellation component still has the potential to progress. We propose an improved neural WPD beamformer called "WPD++" by an enhanced beamforming module in the conventional WPD and a multi-objective loss function for the joint training. The beamforming module is improved by utilizing the spatio-temporal correlation. A multi-objective loss, including the complex spectra domain scale-invariant signal-to-noise ratio (C-Si-SNR) and the magnitude domain mean square error (Mag-MSE), is properly designed to make multiple constraints on the enhanced speech and the desired power of the dry clean signal. Joint training is conducted to optimize the complex-valued mask estimator and the WPD++ beamformer in an end-to-end way. The results show that the proposed WPD++ outperforms several state-of-the-art beamformers on the enhanced speech quality and word error rate (WER) of ASR.