> something like this?
nice. yes, related. in this public situation, was there much accuracy
possible in its recognitions? i don't need it to recognize everything. or
even much. but i do need it to be fairly accurate in what it does recognize.
i don't need 100%. but people have to be able to see that their conversation
and the images are related. the stronger the relation, the better. so the
system just needs to be able to accurately recognize, say, one word every
few seconds and throw out the words it's not real sure about. would be nice
to have quite a large vocab. at least a couple of thousand words. sound
doable within what is currently possible in a public space? i'm thinking it
is possible. people won't have microphone headsets. same sort of situation
as in your piece. only the recognitions will be fed intohttp://vispo.com/dbcinema
you used open source linux voice recognition software?
how was the reaction to the piece?