Network Bending Neural Vocoders

McCallum, Louis and Yee-King, Matthew. 2020. 'Network Bending Neural Vocoders'. In: 4th Workshop on Machine Learning for Creativity and Design at NeurIPS 2020, Vancouver, Canada.. Vancouver, Canada 2020. [Conference or Workshop Item]

No full text available
[img] Text
Network_Bending_DDSP_models-3.pdf - Accepted Version
Permissions: Administrator Access Only

Download (527kB)

Abstract or Description

Network bending [1] aims to elicit interesting creative output from generative
neural networks by applying various transformations to the activations of groups
of network nodes. This paper describes the investigation of how this emerging
technique of ‘network bending’ can be used to provide novel creative control
over sound synthesis networks based on the Magenta DDSP API [2] and how
best to provide access to the resulting sound synthesis neural networks to creative practitioners.

Item Type:

Conference or Workshop Item (Poster)

Departments, Centres and Research Units:



29 October 2020Accepted

Event Location:

Vancouver, Canada

Date range:


Item ID:


Date Deposited:

18 Jan 2021 11:01

Last Modified:

11 Jun 2021 11:35


View statistics for this item...

Edit Record Edit Record (login required)