Gary Schafer, CEO, SIVOX Technologies

Q How is the Sivox RealCall™ product different from other speech-based applications?

A In a nutshell, it's SIVOX RealCall's ability to create a more realistic, interactive experience for the contact center agent.

The heart of the product is its advanced speech recognition engine. It allows SIVOX RealCall to listen to agent responses during the simulation and provide immediate, automatic feedback and remediation when important points are missed via the Real-Time Coach. Learning takes place throughout the simulation, and agents are given the ability to re-try and make corrections before continuing, thereby reinforcing the desired responses. Other simulation products use silence detection that simply waits for the agent's speech to pause before continuing with the script, with no regard for the content of the response. In addition, with other products agents may not know they have made a mistake until they compare their responses to the appropriate ones after the simulation has been completed.

Adding to that experience is SIVOX RealCall's ability to create a simulated replica of the organization's live data systems. This allows agents to practice skills such as data entry and information lookup -- while they're talking -- without putting customer data at risk. Together with the advanced speech recognition engine, the application completely immerses the agent in an interactive simulation that closely mirrors the actual contact center floor.

Q How did you first think of applying speech technology to agent training?

A It started as a custom development project for one of our clients, a leading telecommunications company. As we began designing the application, we realized what a powerful tool speech technology could be, and how it provided for a number of different learning applications. For example, one of our goals was to make sure agents provide a consistent message to customers. This technology is able to help accomplish that goal. Furthermore, the way that we are applying speech technology assures that the important concepts are covered while still leaving room for individuality.

Q Why would someone need to use this application? Please provide actual reasons cited by your customers.

A SIVOX RealCall appeals to organizations on several levels. Some customers want to improve the quality of their customer service. After all, for many of them the contact center agent is the only human interaction customers have with the organization, so they want to be sure the experience is a positive one. Others are looking to reduce training time. One customer has been able to reduce the role-playing time alone by more than 50 percent as a result of using SIVOX RealCall. Another reason is to improve metrics such as reducing call handling time and the number of calls that are escalated to a supervisor. Those organizations are finding that using our products is meeting or even exceeding their expectations.

Finally, some want to protect their investments in training by making sure agents understand what's involved in the job before they hit the sales floor. It can cost anywhere from $6,000 to $10,000 to recruit and train a single agent. Spending that money only to have an agent discover two days into the job that it wasn't what he or she expected is a huge financial risk. SIVOX RealCall provides a realistic preview as to what the agents will be expected to do on-job and allows those who can't handle the job to weed themselves out before our client makes the full investment in the process of hiring and training them.

Q What kind of an effect has using speech recognition had so far on the training of contact center agents, in your experiences? Please provide actual results from deployed systems.

A As I mentioned earlier, one notable effect has been a significant reduction in the time it takes to train an agent. Naturally, that means the cost is cut as well. Just as significantly, our customers are finding that call handling times are going down because agents are better prepared both on the verbal and the data search/entry side.

Ongoing agent training is another area that's been impacted. Keeping agents up to date on the latest marketing offers, for example, is critical to achieving success for a program. SIVOX RealCall has made it possible to create new simulations quickly, without having to involve the IT department, which in turn gets agents up to speed faster.

Here's a great example. In the telecommunications industry, organizations are required by law to tell customers in a certain socioeconomic bracket who are having trouble paying their bills about alternative payment programs. Failure to do so could result in fines - in this case around $17 million. Most agents were unaware of this regulation, and thus weren't making the offer. By including a simulation in the training to cover this situation, our customer was able to greatly increase their success in offering the payment alternative and thus avoid those fines.

Q What is the basis for your learning methodology in relation to training agents?

A As I mentioned earlier, it's essentially to blur the line between training and reality so that when agents hit the sales floor they already feel like veterans. Most contact center agent training is inadequate for preparing agents for the main focus of their jobs - talking to customers on the phone. As a result, once they hit the sales floor there's a lot of on-the-job training going on. While this does provide effective learning, they are gaining experience at the risk of adversely affecting live customers and data.

Our use of speech recognition technology lets them practice their verbal skills on simulated customers in whatever range of experiences the organization thinks they need to have - learn by doing. For example, they can learn how to handle an angry customer in a safe setting. It's the old thought of "If I knew then what I know now." With SIVOX RealCall, they get the opportunity to learn the proper response first, so agents are ready for anything.

The other good thing is they have the opportunity to learn at their own pace, receiving as much (or as little) help as they need. Through the use of speech technology, agents receive immediate, specific feedback on their performance.

Q Describe the process that was used for a recent client in creating a simulation?

A We have designed SIVOX RealCall to make the process of creating simulations easy. If you have a firm understanding of your business, a creative mind, and know how to use a keyboard and a mouse, you can create a new simulation in about four to six hours. An update to an existing simulation can be done in minutes.

We start by meeting with key personnel at the client to understand their specific business challenges, best practices, call flow and other issues. That's always important because all organizations are different and are going through constant change. From there, simulation topics are identified and we begin scripting the actual simulations. SIVOX eLearning experts build the first round using our SimGen tools, a forms-based system that lets experts enter scripts, designate key concepts, add materials to the Knowledge Center, etc. The last step is conducting a knowledge transfer session that allows the client's training developers to begin creating their own simulations from scratch.

Q What effects have you discovered in your deployments that the avatar has had on the overall learning process? What is the basis for having this feature?

A Since we had already created this multi-sensory environment through simulation and speech technology, we wanted to deliver coaching and feedback using the same mechanisms. The avatar personalizes the learning experience and makes it much less intimidating. We wanted agents to get the feeling of working with a helpful, friendly coach or teacher rather than an impersonal machine. Agents have been responding very well to "her," taking the corrections as assistance rather than criticism.

Q Is the spoken output of the simulations TTS, recorded or both? Are there benefits to using one or the other?

A It can be either, but for most purposes we recommend using a recorded voice. The benefit to this method is that you can add emotion to the call. For example, if the caller is supposed to be angry, the current state of TTS cannot convey the required emotion. Using a recorded voice gives the caller more personality, and thus simulates the real world better. Of course, when time is the most important factor, having TTS built in allows simulations to be pushed out quickly. So each method has its advantages.

Q Does your system support multiple languages or only English?

A SIVOX RealCall currently supports fifteen different languages.

Q Where do you see the technology taking you in the future?

A There are a great many possibilities. One of the immediate areas of focus is using our speech recognition technology for agent screening before hiring. In addition to the other tests the HR department is using, they would be able to put a candidate into a simulated contact center environment to see if they have the right skill sets and personality for that type of work.

Further on the horizon, the technology promises possibilities such as verbal branching based on emotion detection. If the agent becomes flustered, nervous, angry, confused, etc. the simulated customer responds accordingly. So even if the words are correct, the system provides remediation on the tone of voice - something a live coach has to do currently.

We are also excited at the prospect of new TTS technology. As I mentioned earlier, we prefer the use of recorded voice in order to provide the theatrical element to the simulation. Future technology could provide the ability to generate emotion-driven speech on-the-fly which would further enhance the experience.

As advanced as SIVOX RealCall is, we've really just scratched the surface of what we'll be able to do with speech recognition technology at the core. It's an exciting time to be involved in this industry.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues