Voice recognition for computers is a nice goal, but I think we might be mistaking the forest for the trees. What we really want is for the computer to take dictation. Transcripts of speech are rough, mean things, with lots of umms and ahhs. They make the speaker sound dumb.
Dictation, on the other hand, is cleaned up speech, turning “what you said” into “what you meant”, via the magic of another person’s brain applied to the problem. I want my computers to take dictation. I can think really well while speaking, but not nearly so well while typing. My typing speech suck compared to my spoken speech. But if I had some sort of agent following me around and transcribing what I say and then translating it from spoken english into written english, and then I can look over the paper for logical inconsistencies and areas of vagueness, my output would go through the roof!
Perhaps the hard part truly is refinement, and having a bunch of ideas that you can pontificate about is trivial. On the other hand, perhaps I should follow the advice that I read in the New York Times – they were interviewing a smoker and drinker who said that he started smoking on the advice of his professors, because the substances would slow his brain down to the speed of his fingers. But that sucks. I don’t want to be dependent on slow fingers. But it’s better than having so many thoughts racing ahead of me that I get lost in midsentence.
Perhaps writers are alcoholics because that’s the only way they can slow their thought process down enough to actually go about the nuts and bolts of writing a story instead of inventing new plots. Can we check out the alcoholism rate among writers as technology shifted from quill to pen to typewriter to word processor? If I’m right, then alcoholism should be dropping as more and more people stop being able to out-think the medium.
