For slow typists like me, speech recognition software has always been an alluring beacon to what I thought would be workplace efficiency. Instead of casting off my qwerty shackles and doubling my words per minute rate, though, I’ve instead wasted a lot of time playing around with junky software that I’ve never gotten to work. (You may have heard me whine about this before…)
ComputerWorld’s Lamont Wood gives a state-of-the-industry report that provides both good news and bad news for folks who are looking to switch a lot of their typing to dictating.
First, the bad news: The best software still needs several hours of training to be effective enough to not be annoying. (If I were willing to put in the time to train my software, I probably would have also taken the time to learn how to type without looking at my keyboard!) Also, there’s no real competition in the marketplace. Only Dragon NaturallySpeaking has made any major advances in the arena, and they are so far out in front at this point that it seems unlikely they’re feeling the competitive push.
Now the good news: Dragon NaturallySpeaking is getting better, perhaps because it realizes it needs to compete against keyboards instead of other voice recognition systems. Wood found out that once he put in the time to train his software, he could dictate 160 wpm with 99% accuracy. (Here’s a video.)
If DNS could come up with a program that learns passively, perhaps I’ll be ready to take the plunge. But right now, figuring out voice recognition software seems like the buying a PDA right before the BlackBerry came out: It’s cool, but it’s a little bit more trouble than it’s worth.
