Audio Texture Synthesis with Scattering Moments

TL;DR

This paper presents a novel audio texture synthesis method using scattering moments, enabling efficient and high-quality synthesis with fewer coefficients than existing techniques.

Contribution

The paper introduces a scattering moments-based algorithm for audio texture synthesis, demonstrating improved efficiency and synthesis quality over prior methods.

Findings

Effective synthesis of audio textures with fewer coefficients

Scattering moments provide robust representations of stationary processes

Gradient descent successfully reconstructs audio signals from scattering moments

Abstract

We introduce an audio texture synthesis algorithm based on scattering moments. A scattering transform is computed by iteratively decomposing a signal with complex wavelet filter banks and computing their amplitude envelop. Scattering moments provide general representations of stationary processes computed as expected values of scattering coefficients. They are estimated with low variance estimators from single realizations. Audio signals having prescribed scattering moments are synthesized with a gradient descent algorithms. Audio synthesis examples show that scattering representation provide good synthesis of audio textures with much fewer coefficients than the state of the art.