BY MARK LEE

The Windows operating system family — Windows NT Server in particular — has become a very important software platform for traditional and Internet telephony over the past few years in client PCs, servers, and embedded systems. The Windows Telephony Applications Programming Interface (TAPI), is one of the important foundation services in the operating system, providing this extensibility for telephony solutions. Microsoft is stepping up its investment in its telephony platform services in Windows 2000, and the new operating system family promises to deliver capabilities to further the cause of open telephony and unified voice and data networks.

Great Solutions Need Great Plumbing
Let’s face it — few people aside from developers really care a lot about the low-level API plumbing stuff. They just want their phone systems and applications to work and deliver value. APIs make it easier for software and hardware made by different vendors to interoperate with one another, because they provide the ground rules for various functions and services. The type of developer plumbing that is used can have a dramatic affect on how long it takes and on how easy it is to develop, deploy, and manage telephony solutions. Plumbing also affects the upfront and recurring maintenance cost of these solutions.

When uniform plumbing is embraced by an entire industry, great economies and efficiencies are enabled, fostering innovation, customer choice, and industry opportunity. Microsoft advocates the use of standardized plumbing, such as the Windows Telephony or NetMeeting APIs, to help developers concentrate on developing great solutions, rather than spending valuable time just getting applications and hardware devices to talk to each other.

Current Platform
TAPI and the Windows NetMeeting APIs are telephony foundations today. TAPI enables software applications to integrate with an organization’s phone system by providing programming “hooks” for call control today. The term call control refers to enabling a software application on a client or server device to automate the way a phone system routes, answers, forwards, or otherwise processes an incoming or outgoing call.

TAPI today allows any type of call model so customers have flexibility in how they configure their phone systems and data networks. It provides a software application and hardware driver model, so both software and hardware vendors can participate. Software vendors telephony-enable their applications with TAPI, and hardware vendors write TAPI service provider software — think of this as code that is somewhat akin to a modem driver in that it allows the phone system to communicate with the Windows operating system. TAPI also enables client/server telephony with several call center features exposed. This is critical in allowing medium to large organizations to economically provide computer telephony integration for hundreds or thousands of users, without requiring special hardware at each desktop.

Windows NetMeeting has become a widely used IP telephony conferencing and collaboration product. It has also become part of the platform due to the NetMeeting Software Development Kit that is available. The SDK allows vendors to integrate their products with NetMeeting or incorporate NetMeeting into their products, taking advantage of NetMeeting’s support for standards such as H.323, T.120, and so on.

THE NEW TAPI
Microsoft is expanding its telephony support with Windows 2000 to make it even easier to connect, automate, and integrate telephony to a client or server system or application. Windows 2000 will include a new version of TAPI — version 3.0; Windows NetMeeting 3; a new Phone Dialer application that exploits some of the new TAPI 3.0 features, including IP Multicast; and Microsoft Web Telephony Engine, something brand new in the Windows 2000 Platform SDK that allows you to use Internet technologies to develop and use telephony applications. TAPI 3.0 benefits developers and customers because it:

• Allows more people to develop telephony solutions in the programming language of their choice. While previous versions of TAPI have been C-language APIs, requiring the use of C++ for development, TAPI 3 is a Component Object Model-Based API. COM technology allows development in Java, Visual Basic, VB Script, or other languages.

• Supports Internet telephony and traditional telephony.

• Expands programming access to media streaming as well as to call control. This gives the developer more standardized control over the actual voice or other media call content — not just the logistics information about the call. TAPI 3.0 integrates with Microsoft’s DirectShow technology, which is particularly useful for client applications but also supports server applications.

• Provides built-in H.323 and IP multicast standards support that is usable by other vendors. Microsoft includes a TAPI service provider and a TAPI media stream provider for both of these important Internet telephony protocols. Now developers won’t have to build or license their own H.323 or IP multicast stacks, so it should make these solutions easier, faster, and less expensive to create.

• Integrates with other OS components and services for manageability and performance. For example, Windows 2000 will provide extensive Quality of Service (QoS) features for a variety of networks — the resource reservation protocol (RSVP) for routed networks, 802.1p for LAN-switched networks, and differentiated services for public carrier use. TAPI 3.0 supports all of these and is administered within the Windows active directory for policy-based networking.

• Adds expanded call center capabilities for routing, queuing, and management of calls for better customer service.

Microsoft also will make TAPI 3.0 available for Windows 98 customers a few months after the launch of Windows 2000. This should increase the number of users who can take advantage of TAPI 3.0, thus expanding the addressable market that vendors can pursue with their TAPI 3.0-based solutions.

The new Windows Phone Dialer application was written for Microsoft by Active Voice Corporation, a company that offers several innovative messaging and call processing solutions. The new Phone Dialer complements Windows NetMeeting 3 by providing a consistent user experience for H.323 point-to-point calls, IP multicast calls, and traditional telephone calls. It uses the built-in H.323 and IP multicast support in TAPI 3.0, and it controls any type of business or residential telephone set that has a TAPI service provider.

CT MEDIA AND TAPI
Microsoft’s recent announcement of its Dialogic CT Media license agreement created quite a stir. The news also created some confusion about Microsoft’s telephony API direction. Skeptics have speculated that licensing this Dialogic (now Intel) technology signaled some sort of abandonment of TAPI looking ahead. That speculation is incorrect.

TAPI and the CT Media software are actually very complementary. TAPI strengths lie in call control, and in gaining access to media streaming resources. The media support TAPI 3.0 exposes today is based on WAV and Direct Show technologies. CT Media server software enables extensive media control over resources on specialized telephony hardware, such as Digital Signal Processors (DSPs), players, recorders, telephony tone detectors and generators, speech processing, and so on. These are media types that WAV and Direct Show do not support today, and they are critical for building complete phone systems or larger voice processing systems on servers or embedded systems. CT Media, which supports the Enterprise Computer Telephony Forum’s (ECTF) S.100 spec, provides very limited call control capability, so it is dependent on other call control APIs, and is ideally suited to work with TAPI.

Microsoft’s intent is to make TAPI the very best, most comprehensive API for call control, while we further enhance its media-handling capabilities to enable rich host-based voice processing apps and to enable hardware-assisted media processing.

NETTING IT OUT
As you can see, there is much that can be done with this platform today, and the future holds even greater promise for new-generation telephony solutions. Vendors can certainly develop to the TAPI and Windows NetMeeting APIs now in Windows NT 4.0 and Windows 98, and make these solutions available to the massive installed base of users of these operating systems. In addition, with Windows 2000’s imminent availability, vendors would be well-advised to take advantage of all the new communications resources available in this platform, so that they can be part of the wave of Windows 2000 adoption that is expected as that product launches.

Mark Lee is Product Manager, Windows Telephony, for Microsoft Corporation. Microsoft products include operating systems for personal computers, server applications for client/server environments, business and consumer productivity applications, interactive media programs, and Internet platform and development tools. Microsoft also offers online services, sells personal computer books and input devices, and researches and develops advanced technology software products.

Microsoft's Web Telephony Engine is one of the new highlights of the upcoming Windows 2000 operating system. It will ship as a royalty-free, redistributable executable in the Microsoft Platform SDK for Windows 2000. It will be covered by the standard Platform SDK license agreement, so organizations will be able to package the Web Telephony Engine code with their solutions, based on Web Telephony Engine.

This open, easy-to-use environment enables Internet technologies and standards to be used to create and run telephony applications. The new service extends the value of existing Web standards, tools, and skills to include voice application support. It also allows developers to use HTML, active server pages, and standard Web authoring tools to develop telephony applications for rapid productivity.

The Web Telephony Engine enables new user experiences and services, and this environment makes it easy to create dual-access Web sites — sites that can be “browsed” via a standard Web browser, or navigated via the telephone using speech or dial-pad navigation commands. The service runs on standard Web servers, and integrates nicely with TAPI 3.0 and Microsoft’s Speech API for even faster, more complete development efforts.

Microsoft’s Web Telephony Engine efforts contributed to the formation of a new working group within the World Wide Web Consortium (W3C) focused on voice browsers. As this is written, more than 70 developers already have been working with early versions of the Web Telephony Engine software to build interactive voice response, messaging, and a number or other communications solutions. It will be interesting to see what sorts of creative solutions result from this new kind of integration of telephony and Internet technologies in the coming months and years.