Using Random Codebooks for Audio Neural AutoEncoders

TL;DR

This paper introduces a novel approach for audio compression using neural autoencoders with random codebooks for discrete representation, demonstrating promising results in audio reconstruction tasks.

Contribution

The paper proposes a new method of building neural discrete representations with random codebooks, advancing audio compression techniques.

Findings

Effective audio reconstruction with random codebooks

Potential for improved data representation in neural autoencoders

Demonstrated advantages over traditional quantization methods

Abstract

Latent representation learning has been an active field of study for decades in numerous applications. Inspired among others by the tokenization from Natural Language Processing and motivated by the research of a simple data representation, recent works have introduced a quantization step into the feature extraction. In this work, we propose a novel strategy to build the neural discrete representation by means of random codebooks. These codebooks are obtained by randomly sampling a large, predefined fixed codebook. We experimentally show the merits and potential of our approach in a task of audio compression and reconstruction.