English

Sounderfeit: Cloning a Physical Model using a Conditional Adversarial Autoencoder

Sound 2018-06-27 v1 Machine Learning Neural and Evolutionary Computing Audio and Speech Processing

Abstract

An adversarial autoencoder conditioned on known parameters of a physical modeling bowed string synthesizer is evaluated for use in parameter estimation and resynthesis tasks. Latent dimensions are provided to capture variance not explained by the conditional parameters. Results are compared with and without the adversarial training, and a system capable of "copying" a given parameter-signal bidirectional relationship is examined. A real-time synthesis system built on a generative, conditioned and regularized neural network is presented, allowing to construct engaging sound synthesizers based purely on recorded data.

Keywords

Cite

@article{arxiv.1806.09617,
  title  = {Sounderfeit: Cloning a Physical Model using a Conditional Adversarial Autoencoder},
  author = {Stephen Sinclair},
  journal= {arXiv preprint arXiv:1806.09617},
  year   = {2018}
}

Comments

Extended conference paper published as article in Brazilian open-access journal Musica Hodie. 17 pages, 10 figures. ISSN 1676-3939. Dispon\'ivel em: https://www.revistas.ufg.br/musica/article/view/53570. arXiv admin note: substantial text overlap with arXiv:1802.08008

R2 v1 2026-06-23T02:41:10.054Z