Voice applications: a reality: tiny screens? Elusive connections? Voice-enable your Web applications so mobile users can access them via phone. It's easier than you think--as long as you take the time to properly plan how you'll integrate Web and voice technologies.

"PRESS 1 FOR ..." You name it. Whether they're interactive voice response (IVR) or speech-based, telephony-based systems are everywhere. U.S. companies provide their customers with an estimated 50 billion minutes per year of service using touchtone-based IVR systems. This translates to more than $3 billion a year for these systems and the infrastructure to support them.

Similarly, virtually every business today has implemented Web-based systems that provide a far richer set of services and capabilities enabling extensive and inexpensive self-service. But, when it comes to mobile customers and users, Web-based systems aren't always the right answer. Everything from screen size, difficulty entering text data, and the "hands-on/eyes-on" mode of operation create challenges to when and how a customers and employees can use mobile devices to access Web-based services.

Although both Web- and voice-based systems are essential to the mobile economy, their evolution has unfortunately been independent. Until recently, the technology to couple the development of these systems and reuse business logic and data assets hasn't been in place. As a result, enterprises have to deal with high costs and inconsistent service quality between the two channels. This article explores the relationship between Web and voice systems, and shows the possibilities for reducing expenses and the user frustration that comes from treating voice and Web applications as separate entities.

When does voice make sense?

Whether you're considering augmenting Web applications or replacing touch-tone telephony applications, there are several key "qualifying factors" that can help you determine if voice makes sense in your enterprise:

* Do you need to reach customers, partners, or associates when Web or other visual access (including WAP, PDA, and smartphone visual access) just "isn't right," but the voice access via a phone is?

* Are your mobile customers or workers seeking hands-free, eyes-free access? Perhaps they're part of a demographic that's less comfortable with or adept at using the Web, or they're in locations where Internet connectivity isn't easily available.

* Does your enterprise manage call centers with live agents (in-sourced or out-sourced) that result in a high cost-per-transaction?

* Are customers and employees dissatisfied with hard-to-navigate, touch-tone applications?

Here are some of the benefits of voice-based applications:

* Voice access to selected applications gives customers and employees anytime, anywhere access via any phone.

* Voice applications dramatically reduce customer and employee service cost when compared to using live agents.

* Many touch-tone based applications cause significant customer dissatisfaction and "zeroing out" to live agents, which drives up transaction costs. Voice applications give users convenient, comfortable human-voice interaction with applications to decrease the transaction time and increase user satisfaction.

Serving the Web and voice channels today: separate and unequal

The diagram in figure 1 represents the implementation paradigm for the vast majority of enterprises with existing Web and voice systems.

[FIGURE 1 OMITTED]

In this common scenario, Web applications and voice applications are completely separate application silos: The Web application uses a three-tier model (more on this in the following section); and the voice application is a single application that intermingles the three tiers indiscriminately.

In many cases, this vertical separation of Web from voice creates a situation where the Web side offers a much broader set of applications because of the ability to quickly and cost-effectively build or extend applications. Conversely, the voice side suffers from a closed, proprietary application environment resulting in high costs for application development and deployment. The cost of building voice-oriented applications has discouraged many companies from adding new features or applications on the voice side, creating voice-oriented services distinctly inferior to Web-based services.

Web applications

Web applications have been long implemented using advanced techniques focused on a three-tier model of presentation, business logic, and data. In the three-tier model, you maintain each layer independently (with development independence at each level of the architecture) to simplify systems implementation and maintenance. For example, you call update Web application styles (e.g., graphics, logos, formats, and browser features) without impacting the underlying business logic or data connectivity. Similarly, you can make changes to data or legacy systems with zero customer impact because the presentation layer and business logic remain unchanged. Note that you can deliver the presentation layer of a Web application to a multiplicity of devices--ranging from standard PC browsers to mobile devices including PDAs and smartphones--by appropriately formatting the content for each type of device.

Voice applications

The Web, with its graphical user interfaces (GUIs), has evolved and matured into a highly sophisticated environment offering clear standards and rules for applications development. Voice systems, however, have developed in a fractured, proprietary market, resulting in pockets of proprietary GUI-based tools--you usually can't port applications between IVR systems from different vendors. For traditional IVRs, you can't do it at all. However, some IVRs are now starting to support standards for building applications which, in theory, should make it possible to port applications between platforms. This is still a challenge though because many companies add proprietary extensions that prevent the application from running on other platforms.

Voice systems generally fall two broad categories: Interactive Voice Response (IVR) systems, best known for their menu-driven touch-tone interaction, and standalone speech-based systems, which leverage automatic speech recognition (ASR). Some systems support both speech and touch-tone applications. The application "speaks" to the user using either pre-recorded prompts or by automatically generating speech using text-to-speech (TTS) technology. Modern TTS has advanced to often be virtually indistinguishable from a live person. Interaction with a voice application is through a voice user interface (VUI), similar in concept to a Web GUI. Replacing the keyboard and mouse input in a GUI, VUIs accept spoken or touch-tone input from phone users. While GUIs return information in a visual format, VUIs render information to users via audio output--i.e., recorded prompts and/or dynamically synthesized speech.

Voice application developers use a markup language similar to HTML and XML already used by Web developers. Using the Voice Extensible Markup Language (VoiceXML), developers can create voice applications that incorporate various scripting languages (e.g., Javascript or Perl) and Java controls to provide additional functionality such as database lookup or dynamic content based on user input. VoiceXML, now several years old, is the defacto standard being adopted by vendors and users. Microsoft has also proffered a second voice application standard, Speech Application Language Tags (SALT), now starting to gain traction since the first products supporting it are just becoming available. VoiceXML is the defacto choice for anyone wanting to build a speech application. Unless an enterprise has a huge investment in using .NET and developing under .NET SDK, SALT isn't the best alternative.

Despite the emergence of VoiceXML as a markup language for voice systems, it hasn't completely eased the problems. Voice applications maintain a separate and unequal status when compared to Web applications. Most tools for VoiceXML-based application development reinforce the standalone voice system paradigm--through build-up of independent data, business logic, and presentation layers--in an attempt to be an all-inclusive, "walled garden" development environment. Also, many tools available today don't reinforce a three-tier development model, in fact some actually discourage that model.

Giving voice users equal access to Web content

With a Web extension to voice paradigm, a voice and Web systems architecture is transformed into the scenario illustrated in figure 2. In this architecture, users can create VUIs to existing business logic and data assets within the enterprise, offering new access for employees, customers and business partners without the inherent limitations of visual-only systems and without the expense of separate, independent voice-based systems. The tools to support application development for this type of paradigm are available today from some of the leading companies in the application server space: BEA has a plug-in for its WebLogic Workshop development toolkit from its partner SandCherry; IBM offers a voice toolkit for Web-sphere studio; and Microsoft has a speech application SDK for Visual Studio .NET.

[FIGURE 2 OMITTED]

By leveraging the vast installed base of Web applications, Web infrastructure, and development community, companies can cost-effectively expand the number of applications available to phones using touch-tone or speech. So, not only are voice applications a reality, they're a smart move.

MOBILE BUSINESS BENEFITS

When It comes to mobile customers and users, Web-based systems aren't always the right answer. Everything from screen size, difficulty entering text data, and the "hands-on/eyes-on" mode of operation create challenges to when and how customers and employees can use mobile devices to access Web-based services.

TAKE AWAYS

Here are the key concepts to understand about phone-based delivery of applications:

1. Voice and touch-tone are essential to complement Web-based services in any multi-channel service delivery strategy. The evolving voice services market represents a significant vehicle for mobile operators and companies to contact customers more often and more effectively.

2. Traditional voice application development methodologies are inherently wasteful and expensive because they involve duplicating development time and costs you've already invested in Web-based services.

3. Web-extension solutions for creating voice applications can reduce development costs by up to 80 percent. By extending Web-based applications--including the data, data connectivity, and business logic--to voice, rather than developing voice applications from scratch using proprietary technologies, companies can focus on just creating new voice user interfaces.

4. Using Web extension can drastically improve ROI for both enterprise and service provider applications. Using a Web extension methodology to reduce development and maintenance costs can dramatically shorten payback time for an application, allowing companies to justify a modest investment for speech-enabled services and offering significant dividends for the business.

Michael Sajor, vice president of worldwide business development for SandCherry, has more than 22 years of experience in business development and management of high-tech companies. Most recently, he served as vice president of business development for multimodal platform provider Kirusa, Inc., where he directed the company's marketing, business development, sales, and product management operations. Sajor began his career at AT&T Bell Laboratories and then served in various positions at Lucent Technologies including director of mobile Internet world markets and product realization for the wireless organization and chief information officer for the company's wireless networks group. Michael has extensive experience working with mobile services and platforms including messaging, multimodality, speech, location-based services, presentation technologies, and mediation gateways. http://www.sandcherry.com