I agree with you, Burns, except for the "will always be" part. It will be useless only until someone finally makes an implementation that Doesn't Suck™.
For voice recognition to really change the way we work, it not only needs to be worlds better than today's, it also needs to be coupled with
language recognition. Simply specifying a few speakable phrases is nice, but...it's a shortcut at best and a tedious longcut at worst. I agree, basically more trouble than it's worth, especially when it's as unreliable as it is today.
Voice+language recognition could potentially be used for much more than simply mimicking keystrokes. It could enable you to do things that there is no interface for, or when the actual interface is just so bulky that it takes minutes to use when it should take seconds (like Spotlight, for instance).
But then, I doubt any big multi-national company would want to tackle that, since they'd need to recognize umpteen different languages! Makes me wonder, as a programmer, which human language would lend itself best to computer recognition. I doubt it would be English, since it's such a franken-language. Maybe Greek? *shrug* Anyway....
With every one of these demos, I lose interest the first time I hear the speaker say "comma" or "period". The computer should be able to tell where punctuation goes. It's not like it's random! But without actual language recognition, it's not possible. Of course, it would raise some questions about the effect on personal writing competence (and style) if the computer took charge of most punctuation, but hey, so did spell check.
Even if used just for dictation, good voice recognition would be useful. Most people can speak a lot faster than they can type.
I'm starting to think that by the time anyone actually makes good voice recognition, it'll already be time for neural interfaces.
