RHR-Net: A Residual Hourglass Recurrent Neural Network for Speech Enhancement

TL;DR

This paper introduces RHR-Net, a fully-recurrent hourglass neural network for speech enhancement that processes waveforms directly, capturing long-range dependencies efficiently and outperforming existing methods.

Contribution

The paper presents a novel end-to-end recurrent hourglass architecture with residual connections for waveform-based speech enhancement, addressing limitations of previous models.

Findings

Outperforms state-of-the-art in six evaluation metrics

Efficiently captures long-range temporal dependencies

Reduces features resolution without information loss

Abstract

Most current speech enhancement models use spectrogram features that require an expensive transformation and result in phase information loss. Previous work has overcome these issues by using convolutional networks to learn long-range temporal correlations across high-resolution waveforms. These models, however, are limited by memory-intensive dilated convolution and aliasing artifacts from upsampling. We introduce an end-to-end fully-recurrent hourglass-shaped neural network architecture with residual connections for waveform-based single-channel speech enhancement. Our model can efficiently capture long-range temporal dependencies by reducing the features resolution without information loss. Experimental results show that our model outperforms state-of-the-art approaches in six evaluation metrics.