Microsoft Names Early Adopters Of Speech ServerMicrosoft Names Early Adopters Of Speech Server

Microsoft is moving closer to giving customers the ability to blend speech-enabled Web applications with call-center operations. The company disclosed on Wednesday the names of early adopters of its Speech Server, which has been in beta testing since July and is scheduled for availability next spring.

Businesses participating in a joint development program for Microsoft's speech technology include GMAC Commercial Mortgage, Grange Insurance, JetBlue Airways, Mary Kay, and New York City's Department of Education. Vendors involved include Intel, Intervoice, Maxxar, ScanSoft, Talx, and Tata Consultancy Services.

Microsoft has been working for 11 years to move speech-recognition technology from the lab to real-world applications. In recent years, much of that effort has been focused on call centers, where the company estimates speech-recognition can do some tasks--say, trade a stock or make a reservation--at a tenth the cost of having a human do the same thing. "There's [potentially] a very large return on investment by automating the call center," says Kai-Fu Lee, corporate VP with Microsoft's speech-technologies group.

A software development kit based on Microsoft's Visual Studio .Net programming tools makes it possible to create dialogue flows that give structure to speech-enabled applications. While some self-service apps lend themselves to speech automation, others--technical support, for example--often require human intervention at some point in the process, Lee says.

Microsoft expects speech-enabled interaction eventually to occur across a variety of devices, including multifunction cell phones. In some cases, Lee says, speech will serve as the means of data input from the device, with information being returned in the form of graphics and text. Says Lee, "We believe that's going to be very interesting to people.".