August 18, 1997
The Power of Voice
By Bob O'Donnell
I've always enjoyed the more science fiction-like aspects of computers, particularly
those things, such as wireless peripherals, that make you stop and say "wow." Of
course, as computer technology has improved, the previously science fiction-like elements
have started to become regular parts of everyday computing life.
One of the most exciting of these elements is the capability to talk to your computer
and have it understand you. Simple voice-recognition systems have been available for
years, of course, but they all require very stilted speech to make them work properly. And
even then, the efforts were marginal at best.
This is finally starting to change thanks to the recent introduction of several new
continuous speech dictation systems from Dragon Systems and IBM. I've been using Dragon's
NaturallySpeaking product for a few weeks now, and although the results aren't perfect,
they are fairly impressive. NaturallySpeaking is the first commercial product that lets
you use continuous speech dictation, which basically means you can talk in a normal voice.
If you've ever tried a dictation system before, you know that this is a big deal.
In fact, this change is important enough that I think it will finally open up these
types of products to the general marketplace. Instead of stumbling through awkward
commands and poor recognition, people can actually get some work done with these products.
As proof of concept, I'm writing this column with my voice. Plus, now that the company has
announced price cuts -- the expected street price is supposed to be about $200 for the
Personal Edition -- it should have more luck reaching a wider audience.
Although I'm tempted to say that voice is the interface of the future, I don't think
it's really true. Certainly right now, products such as NaturallySpeaking still require a
great deal of training and other hand-holding that many people won't put up with. But even
once those problems are solved, I've started to realize that voice isn't an ideal choice
for all types of computer interactions or even all types of writing.
NaturallySpeaking's recognition is good enough to allow me to imagine what it would be
like to completely control your computer via voice, and I've started to see how some
things just won't work. I, and other writers I spoke with, find that more thoughtful
writing is harder to do via voice than it is with the keyboard. Whipping out a quick
e-mail or memo, on the other hand, or transcribing existing text are ideal applications.
Of course, for those suffering from carpal tunnel syndrome or who have physical
disabilities, products such as NaturallySpeaking are a godsend. Like pen computing before
it, voice-based computing seems to offer the potential of a whole new way to interact with
your computer. Unlike the failed promises of pen computing, however, I think voice
computing has the potential to really make an impact both on desktop computers, and even
more importantly, on future hand-held devices. We're not quite to Hal yet, but 2001 may
just be the year that truth is again stranger than science fiction.
©
Copyright 1997, by InfoWorld Publishing Corp., a
subsidiary of IDG Communications, Inc. Reprinted from InfoWorld,
155 Bovet Road, San Mateo, CA 94402. Further reproduction is prohibited.
