The Voight-Kampff Test: Boxing match music performance for four Neural Nets

Date

2020-05-14

Advisors

Journal Title

Journal ISSN

ISSN

DOI

Volume Title

Publisher

GitHub

Type

Software

Peer reviewed

Abstract

This album was created with a process of AI and machine learning called Embodied Intelligence. This uses Creative AI datasets and algorithms to make music from the inside. A single AI-bot was developed to control the interaction of this free improvising garage band featuring sax, bass and drums. No human intervention was used, and the tracks are presented almost as they were recorded, with a minute amount of editing and mastering.

Embodied Intelligence differs from other Music-AI approaches insofar that it is not using AI to construct the physical phenomena of music (i.e. the sound wave), nor the meta-workings of music composition (i.e. the organisation and sequencing of music theory). Instead the Creative AI is powering the realtime generation of core ideas and impetus from within the flow of music-making, which intern is translated into sound.

Embodied AI does not have a model of what it wants to sound like, instead the characteristics of each AI-bot is determined by a front-end translation module. The AI uses Rodney Brooks' approach to robots and 'intelligence without representation' to create a bot that is 'coping in a dynamic and changing world' and 'doing something in this world'.

Description

Keywords

music ai, neural nets, composition

Citation

Vear, C., (2020) The Voight-Kampff Test [computer program]. Available at: https://github.com/craigvear/Voight---Kampff_Test

Rights

Research Institute