Speech Technology Takes Limelight
Jennifer Hagendorf & Stephanie Green
Cambridge, Mass. - The past and presumed future of speech recognition came together during IBM
Corp.'s Speech Fair.
The event, held at the Lotus Development Building in Cambridge, featured demonstrations by
vendors and VARs of how companies use IBM's ViaVoice speech-recognition software. IBM
researchers also were on hand with prototypes of future capabilities that will make the technology
more pervasive.
"We're on the cusp of a technology revolution. I can see it. I'm riding it," said James McGill,
president of Voice Recognition Inc., a VAR based in Needham, Mass., specializing in serving the
health-care market.
The VAR deployed voice technology to streamline the dictation of medical records in the radiation
oncology department at Massachusetts General Hospital, Boston. It was a turnkey package that
included ViaVoice software, integration, customization and training for a one-time fee of less than
$3,000 per doctor, McGill said.
In the past, the doctors dictated their notes onto tapes that were sent out for transcription. The
process took up to a week and was marred by a high rate of error. With the new technology,
doctors dictate their notes right into PCs and can edit the text files immediately while the
information is still fresh, said McGill.
Virage Inc., a vendor based in San Mateo, Calif., presented search technology which uses speech
recognition to turn video into a searchable medium. Sold through VARs, VideoLogger 3.0 and
AudioLogger 1.0 applications currently are in enterprise accounts, federal government agencies and
news organizations such as CNN, ABC and the BBC.
CNN runs all news feeds through Virage technology so reporters can easily search through the
video, said Carlos Montalvo, vice president of marketing at Virage. CNN also features the
technology on its Web site, allowing users to search through testimony from the Microsoft Corp.
antitrust trial, he said.
"The video becomes searchable within five seconds of the stream hitting the logger," said
Montalvo.
In the enterprise, customers such as General Motors Corp. use the technology to index videotapes
of customer focus groups or sales training.
Virage currently works with about 25 VARs and integrators worldwide, Montalvo said.
In the future, voice-recognition capabilities will become part of the everyday human experience,
said those demonstrating their technology.
Researchers at IBM, Armonk, N.Y., are developing biometric speech capabilities, called Voice
Print, that will recognize and authenticate users by their voice, enabling secure telephone banking
and commerce without having to ask the user lots of questions, said Stephane Maes, manager of
IBM's Mobile Speech Solutions.
Another IBM research project is Donatello, a personal assistant that understands natural language
and offers remote access via telephone, so users can access their data from the road. Users can
dictate and send E-mail, change calendar appointments and access their address books just by
talking to their PCs.
Copyright ® 1999 CMP Media Inc.
|
