A Recursive Network with Dynamic Attention for Monaural Speech Enhancement

Abstract

A person tends to generate dynamic attention towards speech under complicated environments. Based on this phenomenon, we propose a framework combining dynamic attention and recursive learning together for monaural speech enhancement. Apart from a major noise reduction network, we design a separated sub-network, which adaptively generates the attention distribution to control the information flow throughout the major network. To effectively decrease the number of trainable parameters, recursive learning is introduced, which means that the network is reused for multiple stages, where the intermediate output in each stage is correlated with a memory mechanism. As a result, a more flexible and better estimation can be obtained. We conduct experiments on TIMIT corpus. Experimental results show that the proposed architecture obtains consistently better performance than recent state-of-the-art models in terms of both PESQ and STOI scores.

arXiv comment5 pages, 3 figures

References in corpus (7)

• Adam: A Method for Stochastic Optimization
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Attention U-Net: Learning Where to Look for the Pancreas
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Delving Deeper into Convolutional Networks for Learning Video Representations
• A Time-domain Monaural Speech Enhancement with Feedback Learning
• Convolutional Recurrent Neural Network Based Progressive Learning for Monaural Speech Enhancement

Cited by in corpus (3)

• Two Heads Are Better Than One: A Two-Stage Approach for Monaural Noise Reduction in the Complex Domain
• Dynamic Attention Based Generative Adversarial Network with Phase Post-Processing for Speech Enhancement
• The IOA System for Deep Noise Suppression Challenge using a Framework Combining Dynamic Attention and Recursive Learning