...wherein we give our projections of how speech product vendors can increase profitability...
Beating the Competition
AS A VENDOR you
probably think the other vendors are competing with you in terms
of "accuracy" of recognition in the desktop market.
They probably are.
But because all vendors have been focusing on accuracy in order
to get magazine reviewers to rave, they have short changed an
area where all current products are really miserable and that
is usability. In our review of the major desktop products
for the November
1999 issue of Software Development Magazine, we found an average
of 75 to 100 usability buglets in each product. These bugs have
always existed - most of them back as far as the first shipping
versions. Other reviewers seem to be oblivious to these. So perhaps
they have been unimportant after all. And obviously, these bugs
have not been serious enough to affect sales.
Or have they? At a February 1999 meeting of Intel VARs in New
York City, Intel proudly told of how the new Pentium III would
help support speech technology. Approximately a third of the audience
shook their heads negatively and indicated that they had abandoned
speech technology altogether. Why do you suppose they had that
We believe that the novelty of talking to your computer has worn
off and that this novelty has accounted for sales to date. Future
sales will come from products whose ease of use is superior.
We use these criteria to evaluate speech
Your current desktop product failed these tests.
Contact us if you are seriously
interested in a formal review of any of your products that are
currently under development.
Back to top...
Winning the Linux market
WANT TO SELL 300,000
UNITS of your speech technology? Here's how. Port to the
Linux platform (but in a special way). Why will that work? Because
of the type of person that predominates in that market. They are
early adopter nerds who work fast, think fast and, most importantly,
communicate with each other fast. So if you sell one of these techies,
you've sold 10.
Reaching the corporate in-house market
But there is a packaging trick you must learn to get these bit
diddlers' attention. You won't so easily sell your dictation products,
SDK's and text readers in their current form. Write
us to find out what should be in that package to make it irresistible
to this market.
Back to top...
NOT TRAINED with great emphasis on traditional language skills.
Oh yes, they learn C++ and ADA and such. But we don't mean programming
languages. We mean speaking languages.
In fact, in modern curriculums, computer science majors are
often given the choice between something like, say, Creative Writing
101 and Compiler Writing 101. Keen students of computers that
they are, they choose the compiler writing course. So they wind
up shortchanged in literal language skills.
Putting programmers with this kind of learning deficiency in
charge of dialogue design for a voice control interactive system
is a recipe for disaster. The same holds true for even simple
voice macro design. The programmers will not have "an ear" for
what makes easy dialogues between humans and computers. The results
will be unsatisfactory systems irrespective of the flash of voice
control. And that will lead to dismissal of the technology by
management. Translation: reduced sales for you the vendor.
A speech products vendor wishing to grow this market, therefore
needs to purposefully engage developers, and use that engagement
as vehicle for training programmers' "ears," training
the capacity to understand what is a smooth audio flow of control
back and forth between computer and human. Lernout & Hauspie,
before their demise, made a start with their Flanders Language
Valley. What are you doing?
For our ideas on what should be in such training write
Back to top...