Apple Developing Computer That Can Recognize Sounds
Apple Computer Inc. has unveiled a working prototype of a Macintosh personal computer that appears to greatly advance the capability of computers to recognize and respond to commands in spoken English.
Spokesmen for the company said the device--which was the focus of a great deal of attention this month at the Technology, Entertainment, and Design conference in Monterey, Calif.--is still in development and many years away from commercial availability.
But they also acknowledged that the device has potentially important applications for education, particularly for physically disabled students.
The device is the brainchild of Kai-fu Lee, a speech-recognition specialist at the firm and a former researcher at Carnegie Mellon University in Pittsburgh, a leading research institution in artificial intelligence.
Speaking at the Monterey conference, John Sculley, Apple's chief executive officer, said the company considers the speech-recognition prototype, which Mr. Lee has dubbed "Casper,'' a "major breakthrough.''
If a commercial application is successful, observers noted, it could eventually render obsolete such interface devices as keyboards and even the ubiquitous electronic "mouse.''
A 'Classic Problem'
How to create a machine with the ability to understand speech is a classic problem in the field of artificial intelligence, and numerous efforts have been under way for years to teach a machine to discern the nuances of spoken language.
Some rudimentary steps in this direction have already been taken, and some of Apple's competitors may also be working on similar products in secrecy.
This month, for example, Dragon Systems Inc., the International Business Machines Corporation, and the Okidata Company furnished a voice-recognition machine with a vocabulary of roughly 7,000 words to John Thompson, a North Dakota high-school student whose arms were severed in a farming accident.
The boy is using the machine to do his schoolwork until his arms, which were successfully reattached, are healed.
But existing systems generally permit the machine to respond only to extremely discrete phrases spoken by highly trained operators.
Observers at the Monterey conference, however, including Marvin Minsky, a professor of computer science at the Massachusetts Institute of Technology and a pioneer in the field of artificial intelligence, reportedly were impressed by the new machine's ability to both comprehend and respond to natural language and to a variety of speakers.
Apple spokesmen said the new technology, which previously has been demonstrated in Japan and Southern California, would eventually end up in Macintosh computers.
But, they said, it could also be applied to a new series of "personal electronics''--machines that serve as electronic notebooks or reference works.
They noted, however, that the device employs the costliest memory microchips now available, which may add significantly to the cost of the final product.
At the same conference, it was announced that Joseph Guglielmi, an I.B.M. employee, has been named the chief executive officer of Talent Inc., a joint project undertaken by I.B.M. and Apple to develop an operating system that will run software from both computer makers.
Vol. 11, Issue 26, Page 10