The W3C Recommendations for VoiceXML 2.0 and Speech Recognition Grammar Specification, approved Tuesday, means the prestigious organization believes the technologies are ready to become industry standards for the Web.
VoiceXML will let developers build Web applications that let people interact with a computer using a telephone. For example, a person could dial into a voice server and ask for stock prices of several companies. The server would carry the request to an application running on a different computer, retrieve the information, and deliver it back to the person as an audio message. The communication between machines and the caller is handled via VoiceXML over an Internet-based infrastructure.
Speech Recognition Grammar Specification is the voice-recognition technology that converts a human voice into computer language.
The two technologies, which have been adopted in products available from more than 80 vendors, are an alternative to using proprietary technologies available for the same purpose, said Jim Ferrans, distinguished member of the technical staff at Motorola Inc. The company was on the W3C working group that developed the specifications, along with Hewlett-Packard, Microsoft, IBM, Tellme Networks, and others.
VoiceXML and SRGS are vendor-neutral, which means applications built around the specifications can run on any software platform supporting the technologies.
"You're basically future-proofing your applications," Ferrans said. "You're not tying it to a particular vendor."
In addition, VoiceXML applications run within Internet-based infrastructures used by most companies today, "so the whole cost of deployment goes way down," Ferrans said. The VoiceXML Forum, formed in 1999 by AT&T, IBM, Lucent Technologies, and Motorola, released VoiceXML 1.0 in 2000. The technology was then handed over to the W3C.