Deep Feature Learning for Medical Acoustics

TL;DR

This paper compares learnable and non-learnable frontends in medical acoustics classification tasks, demonstrating that learnable frontends can improve performance but require larger datasets for effective training.

Contribution

It introduces a framework benchmarking learnable frontends like LEAF and nnAudio against Mel-filterbanks in classifying medical sounds, highlighting their advantages and data requirements.

Findings

Learnable frontends improve classification accuracy.

They require larger datasets for optimal performance.

Benchmarking shows trade-offs in parameters and computational resources.

Abstract

The purpose of this paper is to compare different learnable frontends in medical acoustics tasks. A framework has been implemented to classify human respiratory sounds and heartbeats in two categories, i.e. healthy or affected by pathologies. After obtaining two suitable datasets, we proceeded to classify the sounds using two learnable state-of-art frontends -- LEAF and nnAudio -- plus a non-learnable baseline frontend, i.e. Mel-filterbanks. The computed features are then fed into two different CNN models, namely VGG16 and EfficientNet. The frontends are carefully benchmarked in terms of the number of parameters, computational resources, and effectiveness. This work demonstrates how the integration of learnable frontends in neural audio classification systems may improve performance, especially in the field of medical acoustics. However, the usage of such frameworks makes the needed amount of data even larger. Consequently, they are useful if the amount of data available for training is adequately large to assist the feature learning process.