not [Text]

The audio, video, and interactive works in the ongoing series not [TEXT] explore the human compulsion to find recognition in the barely recognizable and patterns and connections in the barely sensible. The work started as an experiment in the deconstruction of sound, exploring how much abstraction can be introduced before the familiar becomes unrecognizable. It has since evolved to include dynamically generated and irreproducible works of spoken text and images that are rooted in random associations but present as cohesive and compelling stories.

The works in not [TEXT] are created using a combination of custom text-processing software written for the works and various real-time AI image generators and speech synthesizers.

Example Works


Still from not [HOWL], 2023

not [MacB], 2003
4-channel audio
continuous

Excerpt from:
not [Howl], 2023
Single channel video with 2-channel sound
23 minutes, 9 seconds

Ephemeral Poetry Generator #2, 2024
Antique radio case, 3-inch display, Raspberry Pi computer, motion detector, custom software

Ephemeral Poetry Generator #3, 2024
Antique radio case, 4-inch round display, Raspberry Pi computer, custom software

not [MacB] is a radio play for six actors. It was written by an algorithm central to the entire not[Text] project that reconstituted the text of Shakespeares’ MacBeth in a manner that retains the syntax of the original but otherwise pays no regard to meaning. The play is read by six AI “actors” and lines are ordered randomly in a continuous loop, with random scene and act demarcations. The play sounds familiar — unmistakably Shakespeare — but presents a challenge to reconcile the echos of the original with the “plot” of the remixed version.

not [MacB] plays through four speakers in the corners of a darkened room.

not [HOWL] starts with an AI reading of a syntactically similar but semantically jumbled version the Alan Ginsburg’s poem Howl, and pairs it with AI generated images prompted by the new text. However, the images in the video are not synchronized to the words used to generate them. The apparent connections are a hoax, but one that seems in our nature to accept. The text is read by an AI actor with no direction other than what the algorithm derives from the text.

Ephemeral Poetry Generators are devices that create new poems and images in real time in response to user input. The poems are spoken only once and never saved (except for the examples here); and the images persist only until the next poem is created. The poems are constructed from the works of other poets, rigidly adhering to their syntax and vocabulary, but with meaning imposed by chance.

So far, not[TEXT] includes generators based on the poetry of Emily Dickinson, Shamus Heaney, Patti Smith, Sylvia Plath, and William Shakespeare, plus a compendium of cowboy poetry, and writings by Susan Sontag and Gilles Grelet.