Aardvark,
I had experience of two sorts a few years ago, and I'll offer one important suggestion.
First, I knew a woman engineer at the office who used either that or Dragon
Naturally Speaking all the time for composition of reports. Her reason was to avoid worsening her carpal tunnel condition so she really had no choice. She felt it worked fine for her purposes once she got it trained and got used to the special spoken commands for inserting punctuation marks, paragraphs breaks, stc.
And, very important, after learning to speak consistently in the measured, even voice with which she had trained it. She remarked that if she had a slight cold, for example, the recognition went way down. On the other hand, she also had a slight lisp and she was impressed that the machine was able to deal with that. And she also remarked that using the software moved the physical stress from her wrist to her vocal cords. The somewhat unnatural way of speaking stressed her throat and she had to take breaks after every hour to two hours or so of continual speaking.
Second, I picked up either
Viva Voce or
Naturally Speaking for my home computer to try it out and learn something about voice recognition, geek that I am. My computer was on the slow and small side for the software, but I got it to work and trained it to a certain degree. It would work fairly well with noticeable time delays because of the slowness of my computer. So if you do get the software make sure you have
lots of speed and available memory, the more the better. The box will tell you how much, but I suspect even more is better. Of course mine was not a fair test, but I learned what is probably obvious, that correction of typos brings overall speed way down, so training and proper speaking to produce excellent recognition performance are what is required for an overall pleasant experience using the software.
However, both of those sets of experience should be tempered by the realization that recognition technology may have come a long way in the past couple years.
So my strong advice is to seek out someone who actually uses a current version of the software and, if possible, see a demonstration of it, or talk to them about it. It may work much better these days than my descriptions imply. Overall, at the time of my experience it could
not be said that it was the same as speaking to the computer in ordinary conversational tones and speech patterns. HAL was not here yet!
If anyone here has any more recent experience that would be extremely valuable to hear.
Hope this helps,
Peder