What if you had a “software bus” for routing text between applications that convert or translate text, so you could easily create meta-applications that perform multiple conversions or transformations in series in real time. (Hal has already built a prototype of this, and we’d love to see anybody in the DPI community experiment with it.)

You might begin with a speech-to-text conversion, then send the text to a translation program (e.g. English to Russian), then translate it back into English, then send it to a text-to-speech conversion program. Along the way, you could display the intermediate text using real-time animation programs, such as Flash.

What’s it really good for? (Artistically or otherwise?)

Not sure, yet. I’m just fascinated by the kinds of transformations and randomization that happen along the way. What happens when you translate a sentence into Russian, and then Japanese, and then Urdu? Does any sense remain, or is it all gibberish? What kinds of mistakes get made in speech-to-text, and what does it sound like as text-to-speech?

In another way, it’s a kind of “echoing” utility, that allows you to re-hear something you say, in many different ways. It would be great to just have something that animated and displayed what you say, as you say it, with a built-in delay and fading away in the visual distance.

Or it could be a feed into a robot, or into a “Eliza”-type AI respondent.

It could be used in performance, or in an installation. It would be interesting to feed in classic and overly-familiar texts, and see what comes out the other end.

Any other ideas? You can contact DPI if you’re interested in playing with it.