Understanding and evaluating the Caller Experience (CE) provided by voice-driven applications is key to measuring whether a system is achieving its business goals. Knowing how callers perceive a system can help guide business and design decisions by pointing out when callers have trouble versus get helped. SpeechCycle will discuss why this is a critical success factor and talk about methods it has pioneered for measuring caller experience in near real time, providing valuable feedback about true application performance. 

This presentation covers an experiment that was conducted to identify what impacts customer satisfaction. Additionally, this talk provides a list of CSAT factors that can be used to determine which enhancements may help or hurt customer satisfaction. A methodology for prioritizing each factor will also be presented. VUI designers and business stakeholders will learn an approach for accurately measuring CSAT, how to choose the best CSAT question(s), and a method for improving CSAT.

Mobile devices provide a personalized portal to a vast amount of searchable information. Although the GUI on mobile devices remains an efficient output modality, it is believed that providing access to data sources using voice recognition and text to speech will significantly enhance accessibility and uptake. We examine the existing starting points that include IVRs and thick/thin graphical applications. We discuss where and how to change applications to support multimodality and consider trade-offs.

The increasing adoption of wireless communications and smartphones creates the conditions for the evolution of traditional self-service customer care toward a multimodal interaction paradigm. The combination of separate GUI and VUI modes of interaction into a multimodal interface (MUI) allows people to take advantage of their complementary nature for an improved user experience. MUI, together with enterprise voice search—the multimodal evolution of natural language call routing—will drive adoption of self-service customer care toward higher levels of automation.

As companies strive for greater customer intimacy through personalized service, customer-based sales, and proactive outreach, what is the role of speech? Hear real-life examples of successful, customer intimate applicationsand critical steps you can take to use speech to engage, not alienate, customers. We’ll cover the best ways to combine speech with live interactions to enhance the customer experience and how your company can leverage the power of speech for higher levels of customer intimacy and loyalty.

Speech applications need to be designed differently for Spanish than for English.  Cultural and language differences, including "Spanglish," affect design requirements and how callers interact with the application. Our usability study of a U.S. English/Spanish application uncovers new issues.  Menu items, backoff strategies, the preference for touchtone versus speech, and even how to choose participants for a bilingual usability test must be considered.  This presentation will provide a stepping stone for creating better Spanish language speech applications.

When Spanish-speaking users call an automated system in the U.S., the resulting conversation might not be entirely in Spanish, particularly when dealing with names of geographical locations. The way an IVR presents such information can help or hinder usability, and it can make a statement about how a company relates to its Hispanic customers. This presentation will examine key factors for making the best possible choice, such as caller background (demographics, caller preferences, acculturation), inherent characteristics of the location names, common usage, and the purpose of the system.

English speakers globally interact with dialogues differently, depending on their country of origin, culture, and cultural norms. When deploying speech applications in a global context, VUI and grammar designers need to have sensitivity to multicultural differences in the manner in which English is spoken. While most VUI designers realize that designing in another language (such as Spanish) is fundamentally different than designing for English, less is known about the differences in how the English language is spoken globally. Dialogue designs and grammars must reflect these differences for real success.

This presentation will provide new independent research on the factors that come into play when Hispanic consumers choose which language they want to use in an IVR. We present preliminary findings from consumer interviews, using anthropological methodology to observe activities in everyday settings to sift out less visible cultural perspectives. This study challenges the assumption that people who are primarily Spanish speakers prefer to use Spanish IVRs and takes into account variables such as time in the U.S., dynamics of bilingual acquisition, dialectical differences, education level, and attitudes toward English and Spanish.

Speech applications are often designed with the underlying assumption that callers are listening to all the prompts throughout the call. Therefore, when callers do get distracted, they are often left without a means of completing their tasks or achieving their goals. This talk will provide examples of distraction and describe its impact on the caller experience. Specific techniques will be shown that can help callers succeed at tasks, even when they have been distracted.

This prescription for improved customer satisfaction and increased call containment focuses on these best practices: caller-authentication solutions that meet HIPAA guidelines and address concerns about personal privacy and security; assistive dialogue design strategies geared toward an aging population; and intelligent use of backend customer information. This presentation examines solutions in real-world healthcare applications such as prescription refill, new member enrollment, and claim status.

Choosing between text-dependent or text-independent verification greatly impacts the caller's enrollment process, performed to effectively capture the caller's voice print, and the authentication process, used to optimally identify the caller afterward. We also consider functionality such as "liveness" testing that can be used to thwart potential recorded attacks, and the simple "single step" authentication and identification process. Each of these aspects will guide your selection between text-dependent and text-independent verification and ensure that your design attains a proper balance between security and convenience.

This presentation will discuss the key factors underlying customer acceptance of voice biometrics. Based on direct evidence from customer testing in Europe and the U.S., it will address the topics of customer trust and the role of brand in acceptance of biometric technology, mapping customer understanding of biometrics to dialogue design principles, customer perception of trade-offs between security and enrollment, and authentication time and effort.

As the speech recognition grammars required by today's applications become increasingly complex, identifying and fixing grammar coverage problems can become quite challenging. Using real-life examples, this session will provide an overview of some of the best practices and techniques for effective speech grammar debugging and coverage analysis. In particular, we will showcase tools no speech scientist should live without: an interactive sentence explorer and a sophisticated, highly customizable sentence generator.

Weighting is an important and often misunderstood tool in the grammar toolbox. By adding weights to SRGS grammars, developers can help automatic speech recognizers distinguish between similar-sounding words, significantly improving the accuracy of grammar-based recognizers. However, the misuse of weights will almost certainly decrease accuracy. This presentation discusses how weights are used in SRGS grammars, outlines strategies for collecting the data necessary to apply correct weighting, details a strategy of testing, and using real-world grammars and application results, demonstrates how weighting can improve the success of a speech application.

SCXML is a new W3C language for describing control flow. This talk will describe SCXML's mechanisms for communicating with external entities. It will cover communicating with other SCXML sessions, loading data from external sources, invoking external services, and sending and receiving external events.

The Extensible MultiModal Annotation specification (EMMA) is a standard of the World Wide Web Consortium for representing user input. While most commonly used for representing the interpretation of spoken input in dialogues, EMMA provides a great deal of additional information that can also be taken advantage of in speech applications. This presentation will discuss various ways in which this additional information can be used.

Many callers dislike listening to text-to-speech because it sounds robotic and unnatural. While TTS engines today are much better than they used to be, the capabilities of the engine itself is only one driver of the overall quality and audibility of text-to-speech. Other drivers, like quality and integrity of words and abbreviations contained in customer information systems, are equally-if not more-important. But these challenges can be addressed and text-to-speech can be made to sound very natural-it just takes time, resources, and a strong commitment to providing an outstanding voice user interface (VUI). Enterprises that devote resources toward analyzing and tuning the text-to-speech component of their applications on a regular basis, will reap the benefits of a better VUI and improved overall caller experience.

Despite all of the best efforts during the initial VUI design and implementation of a speech-enabled IVR application, changes in the business landscape require continuous monitoring of application performance. Without the utilization of continuous advanced tuning, application performance can erode over time. Avoid this pitfall by planning and allocating resources toward regularly scheduled tuning. This session will offer tuning techniques to ensure ongoing improvements to your application and a successful caller experience.

Optimizing IVR performance to meet clients' service level agreements requires analyzing a wide range of data. There are several methods that can be used to evaluate the data in order to provide recommendations with the biggest benefits. This presentation describes a top-down and bottom-up approach that is driven by SLAs and performance data. These techniques are useful for the VUI designer during design phase and the tuning analyst during post-deployment optimization phase.

The voice user interface forms the hub of the design for any IVR system, with every other part of the design needing to support it. This presentation will share design documentation lessons learned from the implementation of a self-service application for a major U.K. financial institution, giving details of the strategies and techniques used for document production and management, what worked, what didn't work, and ideas for future improvements.

This presentation will cover an emerging process for VUI design that uses a combination of hidden data and automatic data extraction to create a comprehensive design document. First, the presentation will address the pros and cons of commonly used VUI design documentation. Next, it will show how to create hidden data in Visio objects, how to display the hidden data through data graphics, and finally, how to extract and format the hidden data.

Live customer service representatives were getting overloaded with routine flight information questions at one of the leading ultra low-cost carriers based in the U.S. The time-consuming task for representatives to provide flight information was automated with the deployment of the OmniVox3D SIP Application Server and Service Delivery Platform from APEX Voice Communications. Callers looking for flight departure and arrival information simply speak the flight number, departure and arrival city, or airport code to receive the requested information. By accessing the same database as customers on the web, OmniVox3D ensures that the same information is made available to customers calling the automated flight information service.

Continental has been a leader in the development and deployment of self-service technologies for many years. Utilizing a multichannel strategy has enabled the adoption of self-service across multiple touch points and customer groups. In 2008, Continental launched its first outbound IVR for flight check-in. While this application offers an experience similar to that of the online or kiosk check-in experience, it has been designed to leverage the unique capabilities of the voice channel and uses a module-based design, which allows new features to be integrated seamlessly. The outbound flight check-in IVR has since been extended to become an inbound IVR, based on its success and modularity.

Creating an interactive voice and video response travel application for an international deployment involves many parties: the IVVR platform provider's R&D team, in-country sales team, and a business partner for local support; the hosting in-country mobile phone carrier; a travel industry company; and the consumer. This talk will cover the challenges in delivering an IVVR solution that can host not only a travel application, but a stream of applications for a range of hosted customers.

This session will focus on various approaches to adding video support to VoiceXML platforms and the pros and cons of each. The talk will discuss aspects of VXML 3 and how it, as well as some supporting standards such as SMIL, is well suited to support video. The session will begin by looking at the user experience needed for various interactions and then consider this when looking at the underlying technology.