Biomimetic Frontend for Differentiable Audio Processing

TL;DR

This paper introduces a differentiable biomimetic audio processing model inspired by human hearing, combining explainability with deep learning to improve efficiency and robustness in audio tasks using limited data.

Contribution

The authors develop a differentiable, biomimetic audio processing model that integrates traditional signal processing with deep learning, enhancing explainability and data efficiency.

Findings

Outperforms black-box models in computational efficiency

Shows increased robustness with limited training data

Effective in classification and enhancement tasks

Abstract

While models in audio and speech processing are becoming deeper and more end-to-end, they as a consequence need expensive training on large data, and are often brittle. We build on a classical model of human hearing and make it differentiable, so that we can combine traditional explainable biomimetic signal processing approaches with deep-learning frameworks. This allows us to arrive at an expressive and explainable model that is easily trained on modest amounts of data. We apply this model to audio processing tasks, including classification and enhancement. Results show that our differentiable model surpasses black-box approaches in terms of computational efficiency and robustness, even with little training data. We also discuss other potential applications.