Microsoft Talks Pretty One Day

Looking to capitalize on the growing speech recognition market, Microsoft in May released a beta version of the third version of its Speech Server product. The product, called Speech Server 2007, includes new functionality to simplify how companies develop and manage speech-enabled and touch tone applications. Speech Server 2007, which will be generally available in the fall, is the first Speech Server release to support Voice Extensible Markup Language (VoiceXML), a World Wide Web Consortium (W3C) standard, as well as the earlier Speech Application Language Tags (SALT). VoiceXML support will let customers write applications within Microsoft Visual Studio 2005 and deploy them alongside existing VoiceXML 2.1-compliant apps on Speech Server 2007, according to the company. Speech Server 2007 introduces a .NET Framework--based application programming interface (API) for low-level access to core Speech Server functions, enabling companies to build applications with programming languages like JavaScript and C#. The release will feature native support for Session Initiation Protocol (SIP) and Real-Time Transport Protocol (RTP) as core communications protocols for out-of-the-box support for IP telephony and interoperability with leading VoIP gateways, according to Microsoft. Speech Server 2007 also supports TDM and hybrid IP/TDM environments. "This is about giving organizations flexibility on the telephony side of things so that if you're TDM today, but looking at going to IP as most companies are, you've got a road map in the product that's going to support your telephony environment no matter what you choose in the future," says Clint Patterson, director of product management for Microsoft Speech Server. The release also features tools to simplify application development and next-gen speech recognition capabilities based on statistical modeling. For instance, the Grammar Design Advisor catches bad grammars before deployment, highlighting semantically confusing phrases, while the Grammar Tuning Advisor clusters unrecognized words, which can help developers effectively tune the app. Another feature, Conversational Understanding, gives organizations the ability to build sophisticated natural speech grammars, enabling callers to speak more naturally. Dialog Workflow Designer, which is based on the Windows Workflow Foundation, provides a drag-and-drop tool for designing applications and call flows, and can initiate other workflows including business rules, back-end operations, or Web services calls. "The end result is a .NET assembly that's a fully functioning IVR application," Patterson says. Speech Server 2007 supports Canadian French, English (U.S. and U.K.), German, and U.S. Spanish. On the voice response side of the house, some of Microsoft's strongest competition comes from contact center players like Avaya, Genesys Telecommunications Laboratories, and Nortel Networks, according to Steve Cramoysan, research director at Gartner. "Microsoft and IBM are very credible players in other areas and for that reason there's every chance that they will become dominant players in [the speech recognition marketplace], but they've still got work to do." On the speech-recognition technology front, the new Nuance Communications, the result of ScanSoft's 2005 acquisition of the former Nuance, is the clear market leader. "We're happy to see continued investment from providers of general IT solutions like Microsoft because it raises awareness in the category," says Peter Mahoney, vice president of worldwide marketing at Nuance. "We think that most large companies will end up implementing solutions with Nuance and our partners because of the deep specialization we have in speech, the thousands of live applications and the mature platforms vs. a 1.0 version of a VoiceXML browser." Cramoysan admits that he's cautious about overstating the importance of speech, maintaining that in the long run all companies will have to grapple with speech technologies and how they deploy them in their businesses. "Speech access to customer service and speech access to business applications will be in every business."