A voice from the near future

Everyone may focus on the Internet as the best way to share information, but the telephone, having a century's head start, remains the most ubiquitous means of communication. Until now, there's been no easy way to connect the two, but recently developed VoiceXML specifications could prove to be the glue.

VoiceXML is a variation of Extensible Markup Language, which serves as something of a universal translator, tagging data so that different computers know how to process or present it. In the case of VoiceXML, it's a matter of translating dialogue between humans and computers.

Applications developed with VoiceXML will open the Internet to people who cannot use a computer keyboard, whether it's someone who is visually impaired, a field worker with only a wireless phone or anyone else needing to do hands-free Web browsing.

The VoiceXML Forum, an industry group, developed the first set of specifications. It handed the work on VoiceXML 2.0 over to the World Wide Web Consortium (W3C) last October.

Other highly touted voice technologies, such as voice over IP, have not produced much change in how the Web is handled, said Jeff Kunins, manager of developer products for Tellme Networks Inc. Such tools focus on how voice calls are carried over a data network, not on how voice and data can be integrated. Also, they require a lot of time and expertise to insert into phone networks.

"But VoiceXML is pretty transparent and very complementary to what's already in place," Kunins said. "It's just an application layer that sits on top of the current phone system."

It is also an easy way to get much more out of an existing Internet infrastructure, said Bill Dykas, strategic alliances manager for IBM Voice Systems and chairman of the VoiceXML Forum. It requires little or no change to an organization's business processes and few added resources, because the already extensive pool of XML programmers can be used to build VoiceXML applications.

The Environmental Protection Agency used VoiceXML to build a Web-based system that provides public access to a database of more than 3,000 Local Emergency Planning Committees.

The biggest part of the project was pulling together the database, which took about three months but only cost around $1,200 for the database and server software. The VoiceXML tools used to develop, debug and deploy the application were free (from Tellme Networks), and the application itself took only hours to build.

"VoiceXML is something that most people don't know of, or they say they don't have their Web content 'XMLized' yet, so they feel they are not ready for it," said Brand Niemann, a computer scientist and self-styled "XML evangelist" at the EPA and a member of the team that developed the EPA database project. "But they need to understand that if you can apply XML to phone traffic, then they can really start to think about providing universal access."

Every government agency should be able to single out at least one service that could be voice-based and, therefore, could be implemented via the Internet using VoiceXML, Niemann said.

The VoiceXML Forum in February announced its support for the W3C's Multimodal Interaction Activity, which is looking into what standards and software will be needed to access Web applications and services using more than one channel at a time. This will allow someone with a wireless handheld device to access Web applications using both voice and a keypad, for example, something that can't be done now.

"That's what will speed these applications to market," Dykas said. "It's the kind of thing people want, and I expect we'll see them being rolled out in the next two or three years."

Robinson is a freelance journalist based in Portland, Ore. He can be reached at hullite@mindspring.com.