Continental Airlines, Target, Geico, and government agencies such as the New York City Education Department use Speech Server in computer systems that let customers speak commands or choose options over the phone to get information. An update to the software, which can cost as much as $18,000 per CPU, arrived last June. Microsoft plans to discontinue Speech Server at the end of next year and make the software part of Office Communications Server 2007, a new product planned for release next summer.
A new roadmap for speech recognition technology gives software developers one set of APIs for connecting to a telephone network, Web conferencing, instant messaging, and processing touch-tone dialing, says Clint Patterson, a product management director at Microsoft. Companies can also add information about whether a PC user is online to speech-driven software apps. For example, Microsoft will demonstrate at a speech technology conference in New York on Tuesday the ability for two people to engage in a speech-enabled instant messaging session, with one user speaking into a phone and another typing messages on a PC. Development of Office Communications Server comes as companies are designing apps that combine multiple modes of communication.
Microsoft's speech software competes with products from IBM and others. Microsoft released an initial beta version of a planned product called Speech Server 2007 in May that added support for a programming language called VoiceXML that has been backed by IBM, AT&T, Lucent Technologies, and Motorola. Microsoft has backed a standard called Speech Application Language Tags for programmers who use its Visual Studio development tools.
Office Communications Server is due next summer and will replace Live Communications Server. The new Office server uses the Session Initiation Protocol for Internet phone calls and conferencing, which is also used by Microsoft's Exchange e-mail server. A second beta version of Office Communications Server is due in the first quarter of next year.
Microsoft is also building new speech recognition technology into its Windows Vista operating system, due early next year. At SpeechTek 2006 in New York this week, Microsoft plans to demonstrate a Vista feature called Windows Speech Recognition, which the company says can let users issue voice commands to PCs, and dictate documents and e-mails. During a demonstration of the feature at a meeting between Microsoft executives and financial analysts last month, the software failed to recognize much of what a Microsoft manager said. Patterson blamed the failure on a bug in the code.