RealTime IT News

Adding SALT to the Mix

While Philips Electronics, the Dutch consumer electronics giant, may have sold off its speech processing technology division, it's not going very far from the coop; the company that bought the division Oct. 7 announced Friday it would join the standards body co-founded by Philips.

ScanSoft bought up the technology created by Philips for 36 million Euros (or roughly $35.9 million USD). ScanSoft, trying to expand on its core assets, wants to incorporate the technology into its own portfolio of speech enabling Web applications.

The company joins the Speech Application Language Tags (SALT) Forum, a group of companies with the goal of putting speech recognition "tags" into software applications, allowing users to render the spoken word into text and vice versa.

SALT, an open standard organization created by Microsoft, Cisco Systems, Intel and Philips, wants to expand on the technology standard already begun at the World Wide Web Consortium (W3C). The technology could allow blind people to "read" a Word document, and cell phone or PDA users can make phone calls or access the Web by voice.

While the technology has potential in the handheld market (imagine speaking into the cell phone instead of using the keypad to type in a URL), Microsoft has big plans to incorporate it into its .Net Web services platform and TabletPC. With a speech recognition standard, Microsoft would be able to bridge the gap between telephone, handheld and PC.

"Open standards are considered a critical component for the broad development, deployment and acceptance of multi-modal applications," Xuedong Huang, .Net technologies group general manager, said of SALT in a statement Friday.

But he concedes the standard will take some time to see the light of day. One of those areas keeping the standard in the dark, from a developer's perspective, is the tough time SALT will have incorporating with W3C's voice XML 2.0 (VXML) specification.

VXML 2.0 was given the seal of approval by the W3C in June, when it was deemed stable and a likely candidate recommendation for acceptance as a standard. But VXML doesn't have very much in common with SALT, outside the fact both use XML to process voice into text.

Getting the two standards to talk to each other won't be easy. Jim Larson, chairman of the W3C voice browser working group and an employee at Intel's architecture lab, told EETimes it would look to incorporate the two into the next version of VXML.

Ron Schmelzer, a senior analyst at Massachusetts-based ZapThink, thinks the idea of merging the two standards together is a good proposition for developers.

"It would be a good idea for SALT to incorporate with VXML, then you don't have to force people to follow two different specs," he said.

But Schmelzer doesn't see the two combining in the near future, however, because the two standards are only superficially alike. This could create problems down the road.

"It's not a matter of building applications that comply with two different XML documents, it's following the directions of two different standards that's the hard part," he said. "Microsoft and the other groups that created SALT probably didn't think it would impact VXML because they were coming at it from a different perspective; they're just trying to get speech into and out of applications."