Tower is an interactive literary art work where the computer listens to and anticipates what is to be said by those interacting with it. A self-learning system, as the inter-actor speaks the computer displays the next words, in the order of frequency in its database. The speaker may or may not use a displayed word. All new word conjunctions are accumulated and added to the corpus. The initial corpus can be formed from any source. For its premiere exhibition as part of Poetry Beyond Text the corpus was formed as a mash-up of Joyce's Ulysses and Homer's Odyssey.

Words uttered by the inter-actor appear in a red spiral at the top of which the inter-actor is located within the virtual reality environment. Wearing a head mounted display the inter-actor has freedom to look wherever they wish in the environment. However, they cannot move. The words predicted by the system appear as white flickering clouds of associated text in and around the spoken words. What emerges is an archeology of speech (actual and alternate) where the trace of what is spoken can be seen strongest amongst all the alternate things that might have been said. The system constructs an evolutionary tree of language, resembling the Tower of Babel.

The sense of an authentic speaker's voice is challenged by the notion that what has been said has been uttered in relation to the proposed speech of the computer (based on statistical calculations). Has the speaker's speech been normalised through this process? Did they say what they intended? How distinct is the instance of their speech? To what degree is what they say solely authored by the speaker and to what degree is it a function of the emerging collective voice modeled in the statistical prediction system?


one high-end gaming PC with Windows 7
two video projectors
high-specification microphone suitable for voice recognition
one Head Mounted Display unit