September 22, 2004

Big Blue turning up volume on speech app development

Author: Jay Lyman

IBM is throwing its voice to the open source community, but developers
say it is no trick as Big Blue looks to bump up development of speech
software and applications with a worthwhile contribution of code.

IBM, which runs speech labs in the U.S., China and Japan, announced at SpeechTEK 2004 in New York last week that it
would be contributing Reusable Dialog Components (RDCs) -- software tags
needed for date, time, currency, location, zip code, and other functions --
worth $10 million to the Apache Software Foundation and open source
community.

'Enlarging' the speech app environment

Big Blue's Program Director of Conversational Access Brian Garr said the
contribution is aimed at enlarging the environment for the use and sale of
speech technology and allowing developers like him to work with the
components common to speech applications. The speech software building
blocks, which are Java Server Page (JSP) tags, are used to allow features
such as automatic service for callers and can be combined for higher levels
of functionality, IBM said.

The company's speech software initiative, which includes a proposal for the Eclipse Foundation to donate markup editors for speech
standards established by the W3C, won the support of two dozen vendors and
speech software makers, including Apptera, AT&T, Avaya, Cisco, Fluency,
Motorola, Nortel, Nuance, Openstream, ScanSoft, Siebel, Telisma, and Voice
Partners.

IBM said despite its potential, speech technology has been held back by
proprietary code in the "speech ecosystem" that is specific to each vendor,
adding that its open speech initiative was aimed at giving developers the
same tools that mainstream developers have through open, standards-based
models and tools such as the RDCs for the Apache Software Foundation.

Garr said IBM was looking to promote the idea of immediate, within-a-year
return on investment in speech software and to turn up the volume on
speech's overall role in applications.

"We felt the answer was to contribute to the open source community and
sign up 20 or so well-known partners so we all agree to work with the same
components, Garr said.

Referring to a sandbox stage of the initiative where people can
experiment with the RDCs, Garr said the Apache and Eclipse parts of the
strategy made sense, calling the speech software tags that were being
contributed "things everybody had in their kitchen that they could play
with."

'Clear benefits' could lead to quick adoption

The technology was demonstrated at SpeechTEK, where rather than having to
write code on how to gather information, demonstrators were able to drag and
drop a credit card RDC into an application, according to Garr. He predicted
an expanded speech marketplace and quick adoption of the RDCs because of
their clear benefits.

Although IBM's contribution last month of its Cloudscape database code to
the Apache Software Foundation and the open source community for the Derby Project was criticized by some as a case of "abandonware" that had little value, the speech
software contribution has not come into question.

Apache Software Foundation Director Sam Ruby, who is also a TSM for IBM,
said the speech software contribution could be significant but will not
have an immediate impact. Still, Ruby said there is a hunger in the voice
and speech software community for the kinds of tools and tags that IBM is
contributing.

Ruby also made it clear that the code contribution would be judged by
developers on its merits, but that it appeared to be a worthwhile donation.
Ruby said in the cases of both the Cloudscape-Derby contribution and now the
RDCs and voice initiative, the projects have been submitted for
consideration with the Apache Foundation.

"Apache hasn't accepted either of them," Ruby said. "If it turns out it's
abandonware, if there's any inclination that's happening, the projects will
simply be sent back. I don't expect that to happen," Ruby added.

Simplying development is the goal

Ken Rehor, VoiceXML Forum founding member, original co-author of the
voice standard, and Vocalocity chief architect, said the RDCs from IBM
could speed and simplify the development of voice applications.

"Vendor support of standards has greatly expanded the adoption of
speech," Rehor wrote NewsForge in an email. "Standards like VoiceXML, SRGS, and SSML
enable interoperability between platform vendors, application developers,
and tool providers. Customers benefit from these broad choices, since they
can pick and choose products and services that best match their specific
requirements."

"IBM's RDCs support these standards, and can help application developers
build applications more quickly with less development effort," Rehor added.

Click Here!