Problems with Touchtone, and a Speech Recognition Remedy

You often hear people complaining about a touchtone system they've used. Sometimes the technology is implemented in such a way that it confuses or angers callers . And since companies typically use these systems as the front line of telephone communication, customers may be forced to endure a poorly designed system that may aid the company in cost-reduction but may frustrate the caller. Other times the technology is being used for a task that it doesn't serve well.

Movie theater information systems are a common touchtone application, and they illustrate the limitations of the technology. Here's how a popular one works.

When filmgoers call in, the application welcomes them and asks them to press "1" if they know the name of the film they want to see, and "2" if they want to browse all films currently showing.

Callers who press "1" are asked to enter the first three letters of the film's title. This is where things get dicey. The application goes on to tell callers that if the first word of the title has fewer than three letters or if any of those first three letters is "Q" or "Z," they should press "1" again. Callers who press "1" then have to listen to the entire list of current films and press yet another key when they hear the one they want. By this point some callers are probably thinking that they could have walked to the corner newsstand, bought the daily paper, and checked the film schedule in less time.

There is nothing catastrophically wrong with this touchtone interaction. It may give people the information they need ”it's just that touchtone isn't very well suited to the job at hand. The application forces callers to accommodate its limitations ”the fact that each button represents more than one letter and the lack of "q" and "z" on most telephone keypads ”rather than making things as easy as possible for them.

A speech-recognition system's ability to ask a direct question and then act on the response enables a much more efficient transaction. What if we tackled the movie information interaction described above with speech-recognition technology? The application would simply say to callers:

"If you know which film you'd like to see, say its name."

The speech-recognition solution is more elegant, more intuitive, and much faster because the user provides the information in a completely natural way. There is no need to learn how to use the system, no need to hear redundant information, and no need to explicitly decline listening to redundant information. With that one concise prompt, the speech-recognition method can handle every caller need after option "1" in the touchtone application.

Speech-recognition systems can also tailor the calling experience to the needs of the user, offering more detailed assistance as needed to novice or infrequent callers ”and ultra -fast, streamlined service to more experienced callers. Even if a failure occurs, the system can resolve the problem elegantly and optimally, because every facet has been designed and crafted for clarity and comprehension at every moment of the conversation.

When a company implements a speech-recognition system, it can take greater advantage of its phone lines to handle customers' needs while satisfying the needs of the company. With a well-designed system, a company can present a single, consistent personality to the customer, instead of the several ”or even several hundred ”personalities of its call center representatives. During the design process, that personality can be precisely tailored to reinforce the company's brand identity and marketing strategy. Speech-recognition systems can also provide a very high quality experience for the consumer, an impact that cannot be overstated.

A well-designed system doesn't just solve business problems or simply adhere to the company's branding strategy ”it can also create a strong connection between the system and the caller, a connection that is felt viscerally. Deeply. One person who reviewed a system I worked on, reflecting on her experience said, "By the end of using this system, you'll want to take the voice to dinner." ^[3] All it did was provide flight arrival and departure information!

^[3] Julie Vallone, Investor's Business Daily , December 23, 1999.

Speech-recognition technology is constantly improving, too. Today's speech-recognition systems can understand more than just one word or phrase at a time. If a system asks a caller, "OK, what's your travel itinerary ?" and the caller responds, "Boston to San Diego, tomorrow at 3 P.M .," chances are good that the system will understand what was spoken. In fact, a typical North American English speech-recognition system is so good that it can match a single spoken utterance against a list of 80,000 items with upwards of 95% accuracy while still being able to respond to the caller with the next question or statement in tenths of a second.