Siamese SIREN: Audio Compression with Implicit Neural Representations

Luca A. Lanzendörfer, Roger Wattenhofer
6/22/2023
cs.SDcs.AIcs.LG

Official code: lucala/siamese-siren (5 stars, MIT, last commit 2024-01-28).

Abstract

Implicit Neural Representations (INRs) have emerged as a promising method for representing diverse data modalities, including 3D shapes, images, and audio. While recent research has demonstrated successful applications of INRs in image and 3D shape compression, their potential for audio compression remains largely unexplored. Motivated by this, we present a preliminary investigation into the use of INRs for audio compression. Our study introduces Siamese SIREN, a novel approach based on the popular SIREN architecture. Our experimental results indicate that Siamese SIREN achieves superior audio reconstruction fidelity while utilizing fewer network parameters compared to previous INR architectures.

View on arXivView PDF

Code Implementations(1)

lucala/siamese-sirenOfficial100%
MIT

Repository containing code for Siamese SIREN: Audio Compression with Implicit Neural Representations. Published as a workshop paper at ICML 2023 neural compression workshop.

50HTML, PythonMay 21, 20232 years agoMIT
audio-compressionaudio-processingimplicit-neural-representationmachine-learningsiamese-neural-network

Similar Papers

Hypernetworks build Implicit Neural Representations of Sounds

84% match1 repo132023

Improving Accuracy and Efficiency of Implicit Neural Representations: Making SIREN a WINNER

82% match2025

High-Fidelity Audio Compression with Improved RVQGAN

80% match1 repo1.9k2023

Modality-Agnostic Variational Compression of Implicit Neural Representations

80% match2023

EnCodecMAE: Leveraging neural codecs for universal audio representation learning

80% match1 repo1012023

As Good as It KAN Get: High-Fidelity Audio Representation

78% match1 repo62025

Cite this paper

@article{lanzendrfer2023siamese,
  title  = {Siamese SIREN: Audio Compression with Implicit Neural Representations},
  author = {Luca A. Lanzendörfer and Roger Wattenhofer},
  year   = {2023},
  eprint = {2306.12957},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2306.12957v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2306.12957.svg)](https://codeofpaper.com/paper/2306.12957)

Discussion