ADL-MVDR: All Deep Learning MVDR Beamformer for Target Speech Separation
Authors:
Top Cited Papers
- 6 June 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 6089-6093
Abstract
Speech separation algorithms are often used to separate the target speech from other interfering sources. However, purely neural network based speech separation systems often cause nonlinear distortion that is harmful for automatic speech recognition (ASR) systems. The conventional mask-based minimum variance distortionless response (MVDR) beamformer can be used to minimize the distortion, but comes with high level of residual noise. Furthermore, the matrix operations (e.g., matrix inversion) involved in the conventional MVDR solution are sometimes numerically unstable when jointly trained with neural networks. In this paper, we propose a novel all deep learning MVDR framework, where the matrix inversion and eigenvalue decomposition are replaced by two recurrent neural networks (RNNs), to resolve both issues at the same time. The proposed method can greatly reduce the residual noise while keeping the target speech undistorted by leveraging on the RNN-predicted frame-wise beamforming weights. The system is evaluated on a Mandarin audio-visual corpus and compared against several state-of-the-art (SOTA) speech separation systems. Experimental results demonstrate the superiority of the proposed method across several objective metrics and ASR accuracy.Keywords
All Related Versions
This publication has 24 references indexed in Scilit:
- On time-frequency mask estimation for MVDR beamforming with application in robust speech recognition2017
- Improved MVDR Beamforming Using Single-Channel Mask Prediction Networks2016
- Neural network based spectral mask estimation for acoustic beamforming2016
- An Integrated Solution for Online Multichannel Noise Tracking and ReductionIEEE Transactions on Audio, Speech, and Language Processing, 2011
- Speech enhancement with multichannel Wiener filter techniques in multimicrophone binaural hearing aidsThe Journal of the Acoustical Society of America, 2009
- Performance measurement in blind audio source separationIEEE Transactions on Audio, Speech, and Language Processing, 2006
- Design and Analysis of a General Recurrent Neural Network Model for Time-Varying Matrix InversionIEEE Transactions on Neural Networks, 2005
- Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs2002
- A recurrent neural network for real-time matrix inversionApplied Mathematics and Computation, 1993
- Beamforming: a versatile approach to spatial filteringIEEE ASSP Magazine, 1988