Basically, what showerst said.
Basically, what showerst said.
The (theoretical) advantage of voice recognition is that you can speak _much_ faster than you can type (except for a few exceptional cases) [1].
On the other hand, I don't think most people would be comfortable with speaking all day either, but things other than voice recognition could still replace the keyboard.
[1] https://secure.wikimedia.org/wikipedia/en/wiki/Words_per_min...
Try listening to http://www.linux.fm/ for more than three minutes and you'll get my point :)
Unless AI really takes leaps I'm sure this will not happen any time soon. Voice is too ambiguous to convert to text with sound only. The machine will really have to understand you.
Current voice recognition is really awkward and frustrating, and it's currently much easier to learn fast typing than fast voice input.
(and in addition to that there are the many practical drawbacks to voice recognition as voice elsewhere in this thread)