Output details

Autocousmatic

The Autocousmatic project investigated challenges and insights in automating the production of acousmatic music, that is, algorithmically generating electroacoustic art music. A computer program of this name was developed which fundamentally incorporates machine listening (the ability of the computer to analyse audio files, ideally in a human-like manner), with critical feedback to the composing process. This differs from previous single-throughput rule-based systems without such introspection, and ensures listening is primary to the computational modelling of compositional activity in electroacoustic music. The design cycle of composition, through multiple stages of consideration of sound processing and placement of materials, is also reflected in the program itself, and is described in detail in the lead article here.

In order to fully elucidate the research involved in the electroacoustic composing machine, this portfolio comprises a lead article:

(a) (2012) Automatic Composition of Electroacoustic Art Music Utilizing Machine Listening, Computer Music Journal 36/3, 8-23,

accompanied by supporting items:

(b) the Autocousmatic software (available bundled as a standalone and as pure SuperCollider code; http://www.dur.ac.uk/nick.collins/autocousmatic.html) and

(c) SoundCloud examples (https://soundcloud.com/autocousmatic). As well as illustrating the output of the electroacoustic composing machine, these examples played a part in the evaluation of the software described in the article. (Evaluation methods included rejections from juried music festival submissions, user feedback on the software, SoundCloud example comments and activity, expert appraisal of outputs from three electroacoustic composers and discussion with respect to the computational creativity literature).