Windows Speech Server, a server product for creating interactive voice response (IVR) applications, is now available. IVR applications typically enable users to input information by speaking or pressing numbers on the telephone keypad and to get information through a combination of prerecorded prompts and computer-generated speech. These applications are particularly valuable for automating call centers, but could also prove valuable for line-of-business applications that service mobile users. Microsoft and its partners hope it can leverage its strengths in developer tools to make Windows the preferred platform for building IVR applications.

Platform for Voice Applications

Formally known as Microsoft Speech Server 2004, Speech Server delivers two primary components for developing IVR applications:

• Speech Engine Services (SES), which performs all text-to-speech and speech recognition functions
• Telephony Application Services (TAS), which receives telephone calls and enables SES to communicate over the telephone network.

A Speech Server application is composed of a set of ASP.NET pages which contain Speech Application Language Tags (SALT) and, optionally, HTML. When the application is accessed via a telephone, TAS acts as an intermediary, interpreting the SALT tags, while SES receives user input through TAS, interprets it, and generates the responses sent back to the user through TAS.

While the telephone is probably the most important client for IVR applications, Speech Server also allows developers to create applications that can be voice-controlled via a Web browser. A mobile sales professional could use a Pocket PC, for example, to access an internal sales tracking application, and then use voice commands to retrieve the specific sales data for a given customer.

Enterprise Edition Enables Distributed Configurations

Speech Server ships in two editions: Standard Edition, which costs US$7,999 per processor (estimated retail), and Enterprise Edition, which costs US$17,999 per processor. The Enterprise Edition supports multitier configurations in which the TAS and SES components run on separate farms of load-balanced servers for higher scalability and availability. With Standard Edition, in contrast, all Speech Server components must be installed on a single server.

Along with Speech Server, Microsoft is making available a free Speech Application SDK that provides ASP.NET Speech Controls that automatically generate the appropriate mixture of HTML and SALT. The SDK also includes specialized voice application tools to help developers define sets of prerecorded prompts and recognized inputs for an application. Several Microsoft partners are offering "starter kits" that provide an evaluation version of Speech Server along with Intel hardware boards and additional software for computer telephony integration.

Resources

Microsoft's speech Web site is www.microsoft.com/speech.

A more detailed description of Speech Server and its goals appeared in "Server to Aid Voice Response Applications" on page 12 of the Jan. 2004 Update.