W3C sets new standards for voice recognition

By Simon Aughton

Posted on 21 Jun 2007 at 11:48

The World Wide Web Consortium (W3C) has released its Recommendation specification for Voice XML 2.1, which is the standard for voice-driven Web applications. VoiceXML is a language that enables computers to understand human speech and is already used to process millions of calls every day.

VoiceXML 2.1 extends the language with eight features including dynamic access to grammars and scripts, detecting during a prompt that the user is barging-in, and processing multiple sets of data from the server in a single access. All VoiceXML 2.0 applications will work under VoiceXML 2.1 without modification.

The Recommendation was welcomed by Max Ball, director of Product Management at Genesys Telecommunications Laboratories, a leading provider of call centre software.

'VoiceXML 2.1 is a significant milestone in establishing a stable and mature standard that enables businesses to deliver a much richer user experience,' he said.

VoiceXML is one component of the W3C Speech Interface Framework that also includes Speech Recognition Grammar Specification (SRGS 1.0), Speech Synthesis Markup Language (SSML 1.0) and the new Semantic Interpretation for Speech Recognition (SISR 1.0) specification.

SISR 1.0 enables developers to extract and translate textual representations of words detected by a speech recognition system, and then structure the results into a format convenient for processing. For example, with SISR, the sentence "I want to fly from London to Brussels" could be converted to a data structure containing "departure: LHR" and "destination: BRU".

The W3C Voice Browser Working Group will now continue with work on VoiceXML 3, a major upgrade that will include a new XML description language, SCXML.