from the New York times

      April 3, 1997

New Software Greatly Advances Computer Dictation

      By STEPHEN C. MILLER

     C ontinuous speech has been the Holy Grail of the voice/speech
     recognition industry for years, and while it hasn't quite been
     achieved yet, Dragon Systems Inc. unveiled a new computer dictation
     system on Wednesday that represents a significant step forward in
     that quest.

     "Everybody knows that we will be able to talk to our computers,"
     the actor Richard Dreyfuss told a news conference in New York
     announcing the new system, known as Naturally Speaking. "But the
     question has always been: when?" Dreyfuss, who identified himself
     as an unpaid, unofficial spokesman for the Newton, Mass. company
     ("I'm just a fan"), implied that the when was now.

     Naturally Speaking is a so-called continuous-speech program,
     meaning that it allows a user to speak normally to the computer and
     have the words appear on the screen. Current speech or
     voice-recognition products require users to talk in what is known
     as discrete speech, meaning that each word must be pronounced
     individually, separated from the next word by a pause.

     Joel Gould, Dragon Systems' engineering manager, gave the most
     compelling demonstration of the product by using it while
     delivering his prepared remarks so that the words he was speaking
     appeared simultaneously on a giant computer screen. The system
     produced very few errors, and the mistakes it did make were easily
     corrected. To prove that it was not just a canned demonstration,
     Gould read from today's front page of The New York Times.
     "Acknowledged" was interpreted as "a knowledge" and Webster
     Hubbell's last name printed out as "howl," but the system generally
     worked well.
       ______________________________________________________________

     [INLINE] It really works. And it gets better the more you use
     it. [INLINE]

     Scott Miller, Dataquest
       ______________________________________________________________


     A new feature of the program is the integration of command and
     dictate modes. Dragon Systems' current product, Dragon Dictate,
     requires the user to switch between command mode (for example, when
     the speaker says "new paragraph and indent") and dictate mode (when
     the speaker says, "The quick brown fox . . .") Not having to change
     modes makes the product easier to use.

     Yet some complications are unavoidable. For example, the user must
     specify all the punctuation, which represents a new skill for most
     people and can be cumbersome.

     Another challenge in designing dictation programs is making them
     speaker independent. A speaker-independent system will understand
     the speech of anyone, but that is very difficult to achieve since
     accents, regional pronunciations and everyday mumbling make for
     significant differences in how people speak.

     Speaker-dependant software makes a user "train" the program to
     understand that particular user's speech patterns. It can take days
     to train some dictation programs to a degree of accuracy that makes
     the program useful. Usually the user has to recite a long list of
     words that represent the program's "vocabulary." But even after
     training, something as simple as a case of the sniffles can make
     that user unintelligible to the software.

     Dragon Systems isn't claiming total speaker independence, but Dr.
     Janet Baker, the company's president and cofounder, said that
     Naturally Speaking significantly reduced training time. "It takes
     about 18 minutes to train the program," Baker said. That amount of
     time, she asserted, is insignificant when compared to how long it
     takes most people to learn to type.

     Scott Miller, a senior industry analyst at Dataquest, a San Jose,
     Calif., research firm, said he was impressed after being given an
     advance peek at the software. "It really works," Miller said. "And
     it gets better the more you use it."
       ______________________________________________________________

     The big unknown, as usual, is what Microsoft is cooking up.
       ______________________________________________________________


     While Miller was quick to acknowledge the product's imperfections,
     he said that it was ahead of most of the competition. He said that
     the major advantage of Naturally Speaking was that it represented
     the first step in getting people past the biggest barrier to using
     computers -- the ability to type. Just about everyone can talk.

     The big unknown in the voice-recognition industry right now is what
     Microsoft Corporation has up its sleeve. The maker of the Windows
     operating systems has long made clear its interest in incorporating
     voice recognition technologies in future products. Several years
     ago, Microsoft licensed some technology from Dragon, but it is
     currently traveling its own road and keeping its plans under wraps.