Much
of my work to date has focused on the relationship between society and the utopian
promise of technology, in particular the desire for seamless interaction between
humans and technology based systems. From a critical point of view this relationship
might be understood as a spatial problem, based on a belief that intelligence
resides within a technological system, rather than as part of a shared, though
often transparent, interaction. It is these interactions, implicit in our relationship
with technology, with which my work is currently concerned.
Recent works have used voice, both computer generated and human, as a medium
to explore these interactions. Seamless Human Computer Interaction (HCI) is
always imagined in Science Fiction to be conducted via the medium of voice.
In reality, the current abilities of Speech Recognition (SR) systems fall way
short of this fantasy. Current SR software tends to work well in noiseless,
context-free environments. But if a particular context is introduced, perception
of the meanings produced by the interaction may change radically.
In the manner of SR software architecture, I have compiled a lexicon of words
based on my son Jake’s pre-language baby talk. I have extracted these
phonetic sound words and their resulting translations from a 7 hour video of
Jake’s day. These sound words and accompanying translations are a guess,
made by me, on the articulation and meaning of Jake’s words based on
the
context in which they were spoken.
In parallel to this approach, I was keen to create a particular tension in
the
way the audiences interacted with the work. The work’s interface design
and functionality have been informed by quickfix language learning resources.
This approach is taken to an extreme by allowing the audience to decide how
long they wish to spend viewing the work. This decision has an explicit effect
on the manner of the work’s final representation.
Kevin Carter
April 2005
www.co-lab.org