DDSP-based Neural Waveform Synthesis of Polyphinic Guitar Performance From String-Wise Midi Input

No Thumbnail Available

Access rights

openAccess
publishedVersion

URL

Journal Title

Journal ISSN

Volume Title

A4 Artikkeli konferenssijulkaisussa

Date

2024

Major/Subject

Mcode

Degree programme

Language

en

Pages

8

Series

Proceedings of the 27th International Conference on Digital Audio Effects (DAFx24), pp. 208-215, Proceedings of the International Conference on Digital Audio Effects, DAFx

Abstract

We explore the use of neural synthesis for acoustic guitar from string-wise MIDI input. We propose four different systems and compare them with both objective metrics and subjective evaluation against natural audio and a sample-based baseline. We iteratively develop these four systems by making various considerations on the architecture and intermediate tasks, such as predicting pitch and loudness control features. We find that formulating the control feature prediction task as a classification task rather than a regression task yields better results. Furthermore, we find that our simplest proposed system, which directly predicts synthesis parameters from MIDI input performs the best out of the four proposed systems. Audio examples and code are available.

Description

Publisher Copyright: Copyright: © 2024 Nicolas Jonason et al.

Keywords

Other note

Citation

Jonason, N, Wang, X, Cooper, E, Juvela, L, Sturm, B L T & Yamagishi, J 2024, DDSP-based Neural Waveform Synthesis of Polyphinic Guitar Performance From String-Wise Midi Input . in Proceedings of the 27th International Conference on Digital Audio Effects (DAFx24) . Proceedings of the International Conference on Digital Audio Effects, DAFx, University of Surrey, pp. 208-215, International Conference on Digital Audio Effects, Guildford, United Kingdom, 03/09/2024 . < https://www.dafx.de/paper-archive/2024/papers/DAFx24_paper_49.pdf >