"one two won too want to"

"one two won too want to" is a collaborative audio/visual installation consisting of a constantly changing representation of the collective audience. Dynamic input from participants is essential to the output.

Moving away from the familiar keyboard and mouse, a microphone feeding into a speech recognition system is used as an alternative means of interaction. While the output echoed by this imperfect technology mimics the sound and pattern of words, more often than not the computer's (mis)interpretation only approximates the user's input. The transformation continues as the user's spoken words undergo random changes, according to the computer program. String representations of the text are repeated, layered and scaled randomly across the screens. No longer legible as text but simply as image, each message takes the form of a head representing the contribution of the individual. As more and more messages are processed, their predecessors are shrunk to single pixels, the universal unit of computer graphics.

The technology behind "one two won too want to" consists of SONIC, a large vocabulary speech recognition system developed by the CSLR, University of Colorado. It uses a subset of a larger vocabulary file, limiting the amount of words recognized by the speech server. A client application developed in Tcl is connected to a Java program. This program distributes the string representation of the spoken words randomly to the multiple flash files acting as visual front-end on the screens. It also triggers a speech synthesizer, FreeTTS, synchronizing the visual display and the audio interpretation of the text.

Group members:

Roisin Lewis - roisinlewis at hotmail.com
Steffen Karl - steffen.karl at gmail.com


The "Honey" Shop
Everybody at the Firestation Artists Studios
Many thanks to our supervisor Feargal Fitzpatrick

project report