Speech Technology Magazine

 

John Cameron, Chairman, Unveil

John Cameron, chairman of Unveil, explains, "While cost reduction is clearly a driving factor in the purchase of speech recognition, we are seeing an increased emphasis on caller satisfaction."
Posted Feb 1, 2005
Page1 of 1
Bookmark and Share

Q. John, yesterday Unveil announced that you were coming on board as the chairman and will be playing an active role in the organization. What changes or new goals do you have in mind for the company?
A. My immediate goal is to continue to strengthen the company's position as one of the more innovative speech 'middleware' companies in the market today. Unveil has always been focused on driving down the cost, complexity and risk of speech to increase market adoption. Part of this strategy involves innovative technology that mines call recordings to simplify and accelerate speech application development and tuning.  But equally important is a more 'common sense' approach to integrating powerful speech applications with customer service processes at run-time. Effectively integrating these two elements - an environment for the rapid creation and tuning of speech systems and a run-time server that executes applications within the modern Web application infrastructure - we feel represents the best way to dramatically broaden the appeal and adoption of speech applications. Unveil's announcement and demonstration at SpeechTEK of the Unveil Conversation Suite - a complete speech application management environment - is our latest step towards this goal. 

Another key goal is to continue to strengthen Unveil's partnerships.  The company has successfully forged alliances with a range of vendors in the call center marketplace, including platform & infrastructure vendors, speech technology players and, most notably, large scale system integrators. We view these partnerships as essential to making speech more accessible to the broader market and certainly critical to our go to market strategy. The ultimate aim here is an ecosystem of vendor relationships that is mutually beneficial.  Our recent partnership with ScanSoft is another healthy step towards this goal. 

Q. How do you see the speech market evolving over the next five years and what will be the factors contributing to its growth?
A. I believe the speech market is poised for robust growth leading to its assimilation into the broader enterprise software market. There are many indicators that point to this trend, and in fact the evolution of the speech market is beginning to mirror that of many other technology markets before it. For example, take the Web. This technology evolved from custom developed software components with a limited, specialized developer community into a variety of distinct, robust markets integrated into the larger enterprise software industry. The arrival of standards, application tools and middleware fueled growth dramatically increased the community of developers and eventually attracted larger enterprise software vendors to the market.

Today, we find speech at the inflection point between specialization and standardization. The arrival and maturation of VoiceXML has lead to stratification of players in the market - from core technology providers through the platforms, applications and services. VoiceXML gives the buyer tremendous flexibility to choose best of breed components from a more competitive marketplace instead of having to choose between a limited set of proprietary all-in-one systems.

But it is the latest trend in the market - the arrival of packaged components, tools and middleware - that shows real progress towards broader market adoption. It signifies that the speech industry is coming to grips with its reliance on professional services to develop, deliver, integrate and maintain speech systems. A horizontal software layer between the application logic and underlying speech technology is the first step towards standardizing and embedding speech expertise in human factors, VUI design and grammar development in repeatable code tailored to a non-speech expert. It is also this layer of the technology stack that will allow businesses to more fully integrate speech into the enterprise IT infrastructure and; therefore, extend business processes with a speech component.

In other markets, the arrival of real market growth attracted the interest of more established software vendors and lead to a rapid consolidation. I expect this will be the near-term trend for speech as well. It is no wonder that companies like Microsoft, IBM and HP have each increased their level of interest and activity in speech over the last few years. I fully expect vendors from the CRM/ERP and application server markets to also enter the market as speech evolves into a logical extension of their foothold in the enterprise.

Q. What do you feel is the biggest obstacle preventing the widespread adoption of speech today? Is the industry on the right path to solving these obstacles with current approaches to speech application development?
A. The biggest single obstacle to the widespread adoption of speech is risk. Speech risks come in many flavors, but almost all are attributable to the cost and complexity of the underlying technology. There is an extremely limited pool of speech developers with the experience and training to work with the software. The development process is iterative and lengthy, contributing to a long time to deliver on the true value for speech. Implementing a speech application thus forces enterprises to rely on external teams of expensive speech consultants, linguists and grammar engineers. This means the call center - a metric-driven, hands-on part of the enterprise - exercises little control over the speech application lifecycle.

There have been several attempts at reducing the cost and complexity of speech, such as pre-packaged applications, VUIs and grammars. But packaged applications have quickly run into market barriers as well. They are too limited for large scale, complex deployments on the one hand. Conversely, packaged applications are still too reliant on professional services for configuration and 'tuning' to be cost-effective for the mid-market. And so, the fundamental problem remains: the market is looking for an easier and more cost-effective way to directly own and manage the development, deployment and tuning of high performance speech applications.

Unveil addresses this problem by providing a complete speech application management environment to call centers. Through this speech 'abstraction layer,' call center developers can build, deploy and maintain speech systems themselves, eliminating the need for outside professional services expertise. Using intuitive tools, call centers can work directly with the underlying technology to incorporate 'best practices' in VUI design into their voice applications. But this is only the beginning. To deal with the 'usability' risk - customer acceptance of the speech self-service system - enterprise organizations cannot be limited to simple speech dialogs and grammars.  The core speech capabilities have evolved from custom-built, fixed dialog systems into much more natural, conversational applications. Today's speech environments must support tools for building and running these high performance applications as well.

Unveil provides software that circumvents the need for speech experts with information extracted from a call center's existing data. We call this Adaptive Learning - the ability to rapidly build robust speech grammars by mining call recordings for examples of how callers really speak. This data driven approach uses example customer interactions - in the form of call recordings - to 'train' the speech application and help produce a call flow UI. This approach makes use of statistical grammars and results in applications that can be developed in half the time as a fixed grammar natural language speech system while providing more overall flexibility and functionality for the caller.

Q. Unveil recently released its Conversation Suite, what is it and what differentiates - for the customer - it from other similar products?
A. To put it simply, Unveil's Conversation Suite offers businesses better, faster, cheaper speech self-service applications. The Conversation Suite is a complete speech application management environment that simplifies the speech application lifecycle for enterprise call centers.  It includes software for developing, deploying and managing advanced VoiceXML-compliant speech applications. The Unveil Conversation Suite is comprised of the Unveil Conversation Studio, an integrated development environment and the Unveil Conversation Manager, a J2EE-based speech server that manages the execution of applications authored in Unveil Conversation Studio.

Unveil's software addresses two competing demands on call centers today: the need to drive down operating cost and the requirement for high quality customer service. The Unveil Conversation Suite helps call centers better balance these demands by combining innovative technology with a common sense approach to speech automation.  Unlike other speech management environments and tools, the Unveil Conversation Suite makes powerful speech applications simple in the following ways:

  1. It accelerates the development and tuning of high performance speech applications.  The Conversation Suite uses a patented technology called Adaptive Learning™ to put control of speech application development in the hands of the call center. Adaptive Learning mines call recordings for examples of how callers really speak. This approach replaces expensive speech consulting and lengthy iterative design with information extracted from a call center's existing data. Upon deployment, the Conversation Suite continues to collect and learn from customer interactions. As a result, call centers can quickly deploy speech systems that get better over time, minimize tuning costs and reduce the total cost of ownership for speech.
  2. It drives higher rates of automation by producing speech applications that are tuned to your callers' natural behavior.   Because they are developed from examples of how callers really talk, speech applications built with the Unveil Conversation Suite understand callers natural, conversational speech input. These speech applications employ more robust, flexible speech grammars that improve the customer experience and drive higher automation rates.
  3. It increases customer satisfaction by intelligently integrating with live agents. Unveil speech applications use a patented dialog management technique called Conversation Assist™ that provides a 'safety net' for speech. When a transaction cannot be fully automated, Conversation Assist enables live agents to quickly assist the speech application in a way that is transparent to the caller. This can significantly drive up speech automation by containing the caller in the speech system, while at the same time ensuring a more satisfying caller experience.


Q. Is the quality of speech apps keeping pace with customer expectations?
A. In general, yes, but lowered customer expectations are more often the reason for this, not necessarily well designed voice user interfaces. While the industry has made enormous strides in recent years on improving the UI for speech systems, many first generation systems can trap callers in frustrating dialogs or present confusing menu choices.  And in general, most speech systems today are designed to limit what the caller can say and quickly force users into restrictive directed dialogs. If a caller says something unexpected or unsupported by the speech grammars, they are typically re-prompted several times to get them 'back on track.' This design principle is fundamentally flawed in that it forces the caller to adapt to the limitations of the speech software, instead of the speech system adapting to the natural behavior of the caller.

Unveil technologies seeks to correct this imbalance with voice application software that: 

  • Lets the caller interact in a way that is comfortable or natural to them
  • Applies speech and natural language processing technologies to determine the intent of the caller
  • Gets help from live agents when the caller's request cannot be automated without necessarily transferring the caller out of the self-service system
  • Dynamically adapts to caller's natural interactions by capturing caller utterances at system runtime and rapidly folding this data back into the speech grammars with easy to use tools.
     
    Q.  What will be the customer benefits of your recent partnership with ScanSoft and how are these benefits enhanced by the partnership?
    A. Our partnership with ScanSoft enables Unveil to deliver powerful speech applications based on ScanSoft technologies and solutions. This gives our customers more flexibility and choices in the selection of 'best of breed' speech components in the development of their speech applications. These applications can be used to automate virtually any customer service transaction, including conversational call routing, information requests and transactions.  As part of this agreement, Unveil and ScanSoft will be working with existing IVR platform partners to support integrated solutions and services. In addition, Unveil's support of ScanSoft's recently announced xHMI initiative ensures interoperability between Unveil Conversation Suite and future improvements to ScanSoft's speech software and related development tools.

Q. Any thoughts that you would like to add?
A. While there are many contributing factors that will ultimately drive the broader market's adoption of speech, I firmly believe that the industry's singular focus on cost reduction needs to change. While cost reduction is clearly a driving factor in the purchase of speech recognition, we are seeing an increased emphasis on caller satisfaction. That is, call centers need to reduce costs, but cannot do it at the expense of customer service. These two factors are forcing enterprises to take a closer look at the perceived economic benefits of speech in relation to its impact on the caller experience.

Speech offers enterprises an opportunity to enrich their customer interactions, provide more universal access to enterprise-wide information and better extend fundamental business processes with a powerful, flexible speech component.  New methodologies for application design and execution necessitate integrating speech into the overall call center workflow. This more 'holistic' approach to speech requires thinking about the technology not simply as a driver of automation, but as a tool that conforms to overall business rules.  When put in this context, speech can be deployed to increase live agent productivity and extend investments in Customer Relationship Management.

This new thinking about speech self service is reflected in Unveil's partnerships and product announcements. It is our intent to provide solutions that not only increase automation, but also to enable businesses to better balance the competing demands of cost-efficiency and quality of service.

Page1 of 1