Name and Address Capture

THE CHALLENGE

Have you ever ordered something over the telephone from one sales agent, then been transferred to another one who collects your name, address and shipping information? You slowly pronounce your name and address but still have to correct the agent when he or she makes a mistake. As frustrating as it is to you, "name and address" capture is equally frustrating to call center personnel, because of the time wasted during these "spelling lessons".

THE SOLUTION

SpeechWorks Int'l, a supplier of automatic speech recognition (ASR) technologies has developed a better and cheaper way. Using ASR technology, SpeechWorks Int'l automates name and address capture by generating vocabularies in near real time using a computer database. It works this way.

A friendly voice prompt asks for your zip code. Next, a continuous-digit recognizer captures the digits. It now "knows" your zip code. The system then accesses a nationwide database containing over 80 million names and addresses, finds all the street names in your zip code, and builds a phonetic vocabulary of all these names.

A voice then asks what street you live on. The recognizer, using the previously generated "street name" vocabulary, matches your street with the selections in the vocabulary. If an adequate match exists, the system generates another phonetic vocabulary containing the names of all the residents on your street.

The voice then asks for your last name. The recognizer matches your last name with the new "names vocabulary" it has just generated. It now has a zip code, a street name, a last name, and a database that relates the three; enough information to make a well-informed selection, even considering the uncertainties inherent in large vocabulary ASR.

Finally, the voice asks you to confirm your name, street, city, state and zip code. If there is a problem or question, you can be transferred to a live agent.

Call centers are pleased with the system because they save a lot of money, and customers are no longer frustrated by agents who have difficulty hearing or spelling.

At present, a name and address capture application is under trial at a service bureau that previously used live agents. Preliminary results suggest that the system reduces per-address capture costs by approximately 70% when compared to live operators.

Click on the icon below for a demo of name and address capture, built into the CT platform.