The authors discuss how to control outputs from deep learning models of text corpora so as to create contemporary poetic works. They assess whether these controls are successful in the immediate sense of creating stylometric distinctiveness. The specific context is the authors’ piece The Character Thinks Ahead (2016–2017); the potential applications are broad.

Issue Section:
Statement
This content is only available as a PDF.
©2018 ISAST
2018
ISAST

Supplementary data

The Character Thinks Ahead

An AV version of the piece, based on a live performance (Portugal July 2017) given by the authors. Note that this is just one instantiation, as from performance to performance there are not only variations in the way the live text is delivered, but also in the real-time sound processing, and the algorithmic generation of text. The middle panel of composed text does not vary between performances. (© Roger Dean and Hazel Smith)

- mp4 file
You do not currently have access to this content.