A convolutional recurrent neural network with attention framework for speech separation in monaural recordings

Guo Xian,Xiaofeng Gong,Chao Sun,Ruisen Luo,Ruijuan Wu,Min Zhang,Qin Yu,Junhong Lu

doi:10.1038/s41598-020-80713-3

Abstract

Most speech separation studies in monaural channel use only a single type of network, and the separation effect is typically not satisfactory, posing difficulties for high quality speech separation. In this study, we propose a convolutional recurrent neural network with an attention (CRNN-A) framework for speech separation, fusing advantages of two networks together. The proposed separation framework uses a convolutional neural network (CNN) as the front-end of a recurrent neural network (RNN), alleviating the problem that a sole RNN cannot effectively learn the necessary features. This framework makes use of the translation invariance provided by CNN to extract information without modifying the original signals. Within the supplemented CNN, two different convolution kernels are designed to capture information in both the time and frequency domains of the input spectrogram. After concatenating the time-domain and the frequency-domain feature maps, the feature information of speech is exploited through consecutive convolutional layers. Finally, the feature map learned from the front-end CNN is combined with the original spectrogram and is sent to the back-end RNN. Further, the attention mechanism is further incorporated, focusing on the relationship among different feature maps. The effectiveness of the proposed method is evaluated on the standard dataset MIR-1K and the results prove that the proposed method outperforms the baseline RNN and other popular speech separation methods, in terms of GNSDR (gloabl normalised source-to-distortion ratio), GSIR (global source-to-interferences ratio), and GSAR (gloabl source-to-artifacts ratio). In summary, the proposed CRNN-A framework can effectively combine the advantages of CNN and RNN, and further optimise the separation performance via the attention mechanism. The proposed framework can shed a new light on speech separation, speech enhancement, and other related fields.

Highlights

Most speech separation studies in monaural channel use only a single type of network, and the separation effect is typically not satisfactory, posing difficulties for high quality speech separation
In order to only compare the effects of different loss functions on separation, the neural network we used in this part does not contain an attention mechanism (i.e. CRNN)
It can be seen from the definition of SIR in Eq (12) that einterf decreases, and the reduction of einterf is greater than etarget

Summary

Introduction

Most speech separation studies in monaural channel use only a single type of network, and the separation effect is typically not satisfactory, posing difficulties for high quality speech separation. The proposed separation framework uses a convolutional neural network (CNN) as the front-end of a recurrent neural network (RNN), alleviating the problem that a sole RNN cannot effectively learn the necessary features. This framework makes use of the translation invariance provided by CNN to extract information without modifying the original signals. Yang et al.[8] considered the singing voice to be a sparse signal, while the accompaniment part can be represented by a low rank. Roux et al.[11] proposed a deep non-negative matrix factorisation (DNMF) separation algorithm and expanded the NMF iteration, improving the accuracy with fewer parameters

Objectives

Methods

Results

Discussion

Conclusion