On Filter Generalization for Music Bandwidth Extension Using Deep Neural Networks

TL;DR

This paper investigates how the choice of low pass filters affects deep neural network performance in music bandwidth extension, proposing a data augmentation method to improve generalization across different filters.

Contribution

It highlights the impact of filter mismatch on neural network performance and introduces a data augmentation strategy to enhance filter generalization in audio bandwidth extension.

Findings

Matching training and testing filters yields up to 7dB SNR improvement.

Filter mismatch can reduce SNR, sometimes below input levels.

Data augmentation with multiple filters improves generalization.

Abstract

In this paper, we address a sub-topic of the broad domain of audio enhancement, namely musical audio bandwidth extension. We formulate the bandwidth extension problem using deep neural networks, where a band-limited signal is provided as input to the network, with the goal of reconstructing a full-bandwidth output. Our main contribution centers on the impact of the choice of low pass filter when training and subsequently testing the network. For two different state of the art deep architectures, ResNet and U-Net, we demonstrate that when the training and testing filters are matched, improvements in signal-to-noise ratio (SNR) of up to 7dB can be obtained. However, when these filters differ, the improvement falls considerably and under some training conditions results in a lower SNR than the band-limited input. To circumvent this apparent overfitting to filter shape, we propose a data augmentation strategy which utilizes multiple low pass filters during training and leads to improved generalization to unseen filtering conditions at test time.