Coalition to Create Speech Standard for Web Services

• By Matt Migliore
• 10/19/2001

The specification, Speech Application Language Tags, will build on existing markup languages such as HTML, xHTML, XML and WML. The development process will be driven by a working group – also announced this week – called the SALT Forum. Its founding members are Cisco, Intel and Microsoft, as well as speech-technology players Comverse Inc., Philips Speech Processing and SpeechWorks International Inc.

The forum is positioning SALT as a platform-independent standard for multimodal and telephony-enabled access to information, applications and Web services from a variety of devices, including PCs, cell phones and personal digital assistants. SALT is intended to allow data input via speech or a keyboard, keypad, mouse or stylus; and production of data as audio, plain text, motion video or graphics, either simultaneously or independently.

In the framework of Web services, SALT is expected to make it easier for developers to present data in a variety of different contexts – a function that has been fairly elusive in the present-day Web application environment.

“Since SALT [will be] a lightweight, add-on tag, you can see how it will plug into a [Web services] environment very nicely,” says Kevin Wiant, vice president of business development for Comverse. Wiant says he sees practical applications of SALT in the form of voice-enabled services for mobile phone users who are driving a car.

Such capabilities have been discussed in the speech-application space for several years now. And standards like Voice XML have emerged in an effort to bring speech-enabled services to users.

Voice XML, which was first introduced in March of 1999 and has achieved a fair amount of acceptance in the industry, is overlapped somewhat by SALT. Both standards offer features for working with data through speech. However, Ken Hyers, a senior analyst with the mobile commerce division of Cahners In-Stat Group, doesn’t see VXML and SALT as competitive standards.

“I get the impression that [the founders] want to bring Voice XML into the [SALT] forum,” says Hyers. Hyers believes the Voice XML and SALT forums will ultimately work together to develop a single standard.

“[Voice XML and SALT] have identified some real problems, in that developers are not building voice into their software,” says Hyers. “I think the reason is that there’s not really one standard to develop around.” Hyers feels that a single standard could emerge from the combination of SALT and Voice XML.

Steve Chambers, vice president of worldwide marketing for SpeechWorks, says the SALT Forum is currently recruiting contributors to help refine the specification. He says, both wire and wireline network operators, application developers, speech technology firms, infrastructure providers, and handset companies are the forum’s key recruiting areas. But he does not say whether the SALT group is pursing a working relationship with the Voice XML Forum.

Even without Voice XML’s support, SALT already figures to gain some solid backing from developers. Microsoft has vowed to create SALT extensions for its Visual Studio .NET and ASP.NET offerings, as well as for the Internet Explorer and Pocket Internet Explorer Web browsers.

“Microsoft is talking about this as part of their Visual Studio .NET platform,” says Chambers. “So, right away, that bring in a whole bunch of developers.”

Still, Hyers believes a common standard is necessary for speech and multimodal technologies to move forward. Furthermore, Hyers says, in order to achieve widespread adoption of multimodal services developers must present data in context-specific ways.

“If I’m in a crowd and I want to know something about anthrax, for example, I don’t want a 3,000-word article read to me,” says Hyers. “I want the key points – an executive summary of some sort.”