Voice Technology

COMMUNICATIONS GUIDE

Voice Technology

Interactive Response and Messaging Products Boost Customer Service

oice response and messaging products are helping agencies handle high-volume phone traffic quickly and efficiently. State-of-the-art systems can answer routine questions while eliminating busy signals, long waits on hold and games of telephone tag. They even can fax useful information to callers.

Voice mail, automated attendants and call-processing devices have replaced human telephone operators in many government offices. These products answer incoming calls and route them to appropriate parties. Air Force hospitals use a Bell Atlantic system that enables patients to leave messages for doctors 24 hours a day. An Active Voice system helps the Small Business Administration provide information on company start-ups, loans and minority programs.

Interactive voice-response technology from companies such as Applied Voice Technology, Maxxar Corp. and Microlog is being used to translate spoken words into digital signals that are processed by PCs. Special software compares the signals to sounds stored in computer databases and makes comparisons until matches are made. The technology is ideal for applications using only a few words.

Applications using continuous speech must rely on voice-recognition technology, which uses mathematical techniques to dissect speech into building blocks that differentiate between various accents and nuances of male and female voices. The IRS is using Periphonics speech-recognition products in its TeleFile system for filing tax returns by phone. And the Social Security Administration uses a voice-recognition system for its hot line.

The Education Department routinely uses interactive voice-response products to automate the collection of delinquent student loans, while the Bureau of Labor Statistics uses them for gathering monthly employment survey data. But voice-recognition technology is far from perfect. Since so many letters and words sound alike but are pronounced differently, users have to pause between words to give computers time for analysis. In addition, background noise can throw off recognition devices.