Industry panel proposes voice standard for Web

A coalition of leading technology companies working on speech technology Monday announced plans to drive a new standard for accessing Web services and other information on the Internet through voice commands.

Speaking at an event on Microsoft's Silicon Valley campus, an industry panel of executives from companies including Cisco, Intel and Microsoft unveiled a proposed standard for developing voice-activated Web applications, called Speech Application Language Tags (SALT).

The technology is intended to make Web services accessible through a telephone or any computing device that can transmit voice. With the technology, these companies say voice will replace the functions of a keyboard, telephone dialpad or handheld stylus.

"Voice is the only modality that spans through these devices," said Kai-Fu Lee, vice president of Microsoft's natural interactive services group.

In addition to Microsoft, Intel and Cisco, founding members of the newly formed SALT Forum include Royal Philips Electronics' speech technology division, Comverse and SpeechWorks International. More companies are expected to join the Forum as the project proceeds, the group said.

SALT will enable companies to build voice commands into their sites, similar to other services already on the market from BeVocal and AOL Time Warner, with its AOL by Phone service. The panel touting the technology said the market for voice-enabled services is poised to boom as devices get smaller and require an easy method for accessing services.

Microsoft said it would incorporate SALT into Visual Studio.Net as the technology evolves so developers can build voice-activated applications with the development software suite. The technology will also be supported in the Internet Explorer Web browser, and will be part of Microsoft's emerging .Net initiative for delivering services across the Internet.

"Speech is, in the future, really the ideal interface for the Web and .Net," Microsoft's Lee said.

SALT will work independent of the operating system and on any computing device that embraces the standard, the companies said. The technology will also be free for use by Web developers.

Although the idea has gained wide industry support, it may slow down the development of another standards group working on a similar technology, called VoiceXML (Voice Extensible Markup Language), said John Shea, director of product marketing at Nuance Communications, a voice recognition company working with Voice XML. Those working on the SALT project say it is expected to be a lightweight alternative to VoiceXML.

"We have these Web sites and content, and with some very simple SALT it is easy to get [voice] ccess to all these sites without them being rebuilt," said Frank Caris, president of Philips Speech Processing North America. Philips will contribute much of its voice recognition research to the group, including technology to localize speech recognition to various languages and dialects.

The standard is based on HTML (HyperText Markup Language), XML (Extensible Markup Language) and other Web development standards. It will be one of many technologies designed to facilitate a widely touted concept of delivering information to users at any time and on any computing device, from a cell phone to handheld. Microsoft, as well as industry rivals Sun Microsystems and Oracle, has long pledged to build software that is delivered as a service across the Internet.

In addition to creating a standard method for sending and receiving voice messages, SALT will allow hardware and software makers to create a standard user interface for hosting new voice services on computing devices. The SALT Forum will work with companies to build a speech interface onto a Web site so it can be accessed by telephone or mobile device.

"The boundary between a computer interaction and a human interaction should be dropped," said Howard Bubb, vice president of the telecommunications and embedded group at Intel.

SpeechWorks International, a provider of speech recognition and voice-enabled applications, currently offers tools to build a speech interface into a Web site. Its customers include Wal-Mart Stores, Amtrak and United Airlines.

"This isn't that futuristic," said Steve Chambers, vice president of worldwide marketing at SpeechWorks. "Companies are creating these applications now."

The initial version of the specification will be available in the first quarter of 2002 and will be submitted to a standards body for review by the middle of the year, the companies said.

The IDG News Service is a Network World affiliate.