HyperSound: Generating Implicit Neural Representations of Audio Signals with Hypernetworks

TL;DR

HyperSound introduces a meta-learning hypernetwork approach to generate implicit neural representations for audio signals, enabling high-quality reconstruction of unseen sounds without training separate models for each sample.

Contribution

It is the first to adapt hypernetwork-based INRs to audio, allowing efficient, sample-independent audio representation and reconstruction.

Findings

Achieves sound wave reconstruction quality comparable to state-of-the-art models.

Eliminates need for training separate models per audio sample.

Demonstrates effectiveness on various audio signals.

Abstract

Implicit neural representations (INRs) are a rapidly growing research field, which provides alternative ways to represent multimedia signals. Recent applications of INRs include image super-resolution, compression of high-dimensional signals, or 3D rendering. However, these solutions usually focus on visual data, and adapting them to the audio domain is not trivial. Moreover, it requires a separately trained model for every data sample. To address this limitation, we propose HyperSound, a meta-learning method leveraging hypernetworks to produce INRs for audio signals unseen at training time. We show that our approach can reconstruct sound waves with quality comparable to other state-of-the-art models.