When US Foodservice was seeking help boosting productivity, voice technology answered the call. As part of a warehouse improvement initiative back in January 2001, the nationwide wholesale food distributor was considering installing a voice technology system in one of its warehouses. At that time, the bidder, Voxware Inc., made a promise, recalls Steve Fasulka, a vice president of operations at US Foodservice's Pittston, Pa., division. "The company promised a 50-percent reduction in mispicks and shorts," Fasulka says. Did it deliver on the promise? "Not only did it achieve that goal," he answers, "but the company exceeded it."

Since then, US Foodservice, which is headquartered in Columbia, Md., has rolled out voice technology—which allows a computer to both recognize spoken words and provide "spoken" instructions—in a half dozen warehouses across the company. Fasulka reports that the biggest productivity increase in his division has resulted from the way warehouse order pickers now handle "catchweights," items like meats that must be pulled out of a case and weighed before shipment. In the past, the warehouse worker had to take precooked hams and turkeys out of a case, weigh them on a scale, record the weight on a slip, and then turn over that information to an office clerk, who keyed in the invoice data. Now, the selector speaks into a headset to record the weight while weighing the food item, and the spoken words are then digitized into text. "We saw a 50-percent increase in productivity in that area," says Fasulka.

US Foodservice is not alone. In the past six months, a number of companies—particularly in the grocery industry—have begun deploying this technology to increase distribution center productivity. Because the worker can communicate with a computer via a headset, his or her hands are free for tasks such as picking or weighing objects while receiving instructions from a computer. "Companies are more serious about this technology in hard times," observes speech industry consultant Judith Markowitz in Chicago. "It enables a person to perform more work single handedly."

Although a few high-profile companies like US Foodservice have installed voice technology in their distribution centers, the equipment is just starting to gain traction in logistics. Warehousing consultants contacted for this article all note that concerns about cost and demonstrated benefits have deterred greater adoption until now.

Although scientists at Bell Labs began working on speech technology prior to World War II, the federal government drove the real advances in this area by funding research during the early Cold War period. As computer manufacturers introduced hardware with higher processing speeds and enhanced memory capacity, companies sprang up to take advantage of that research and develop software that could recognize and reproduce the human voice.

How big is that market today? The investment firm Morgan Keegan & Co. Inc. of Memphis, Tenn., estimates that the three main types of speech technology—speech recognition software, speaker verification software and text-to-speech software—accounted for $153 million in sales in 2000. It expects the market to exceed $1.5 billion by 2005.

Most of that equipment is sold for what are known as telephony applications—those providing directory assistance, automated call direction or unified messages. Vendors of speech recognition and voice synthesis software initially marketed their programs to companies looking to streamline administrative tasks such as secretarial dictation. But today the technology is more apt to be found in customer service call centers, where an automated voice handles telephone inquiries. Automakers have even started deploying the technology in automobile navigation systems.

The development of speech recognition and voice synthesis systems targeted to the warehouse took place about a decade ago. In a typical application, these systems receive instructions from a computer application such as an order management or warehouse management system. The communication system first synthesizes the data into speech, enabling workers equipped with a headset to receive oral instructions from a computer system directing their work. In turn, workers can use a microphone to respond to the computer to note that their task has been completed. In warehouses, such systems have been used for parts inspection, putaway and order selection. The key advantage to this technology is that the warehouse worker's hands are free to lift items or operate equipment as he or she communicates with a computer. "While the next location is voiced to them," says Fasulka of US Foodservice, "our workers can keep their hands on the pallet jacks they're operating."

At the moment, there are two voice-recognition systems vendors active in the distribution and warehousing market—Voxware Inc. of Princeton, N.J., and Vocollect of Pittsburgh. Both sell what are known as dependent voice-recognition systems, meaning that the worker must create a "voice template" so the system can recognize his or her speech patterns to decipher a predefined set of vocabulary words.

A publicly held firm, Voxware reported revenues of $2 million for its last fiscal year, which ended on June 30, 2001. But revenue for this year has picked up, says Ken Finkel, Voxware's vice president of business development. Finkel reports that his company's revenues for the first nine months of the current fiscal year are double what they were for the same period a year ago, a sum of $1.14 million. "We have exceeded our expectations. We're installing multiple sites," he says, "and we're in the field installing all the time."

Finkel attributes the rise in sales during the first six months of this year to the technology's demonstrated ability to increase accuracy and productivity within distribution. "The payback [time] on most systems is less than a year," says Finkel. "These systems can be justified in this economy." He says Voxware's units cost between $4,000 and $5,000 per user, an expense that includes the voice application itself and software for the server.

Rival Vocollect also reports that there has been an upswing in sales and interest in its technology for the first six months of this year. The company's vice president of marketing, Larry Sweeney, says Vocollect has sold more than 12,000 terminals to date. This year, he expects that the privately held firm will exceed $20 million in revenue, a 70-percent boost in sales. "Quick payback is driving this market," he says. "[The main selling points are] accuracy and productivity."

Sweeney says Vocollect's units go for $4,000 per worker, a figure that includes the headsets, training employees to use the technology, and the software. He adds that customers usually receive a payback on their investment in six to nine months.

Vocollect has concentrated its sales efforts for the most part on the grocery industry, which has been receptive to this technology. "The grocery distribution folks are pretty advanced," says Sweeney. "They have led the charge over time. They are looking to get the extra percentage point out of their operations."

With its thin profit margins, the grocery industry is deploying voice technology in hopes that it will boost worker productivity in the distribution center. Sweeney claims that his company's voice technology typically brings about an 80-percent reduction in picking errors while boosting worker productivity by 15 to 20 percent.

Although Vocollect and Voxware have both reported rising sales for the first part of this year, industry experts note that the warehousing market remains wide open. "The penetration in warehouses is still minimal," says Bill Meisel, editor of SpeechRecognition Update, a Tarzana, Calif.-based newsletter covering the voice technology market.

Limitations inherent in the current generation of technology have made many companies hesitant to embrace it, adds consultant Robert Mann in the Atlanta office of Accenture. He notes that the speed of a voice technology system often depends on the amount of vocabulary it's expected to recognize. "If you find an application using voice, it tends to be limited to the words "yes" and "no" and the numbers zero to nine," he says. "If you start expecting more general input, then it gets dicey in terms of its effectiveness."

In addition, background noise can interfere with the operation of speech recognition systems deployed in industrial environments. "Background noise makes it difficult for these systems to recognize the worker's voice, although there have been great strides in the last two years," says John Kelly, who's the publisher and editor in chief of Speech Technology magazine in Lexington, Ky.

Furthermore, companies often find they must install a radio-frequency network along with a voice system if they want to ensure real-time communication with the workforce. The additional expense for implementing a radio-frequency network often causes companies to think twice about going forward with such a project.

Despite those obstacles, voice technology has started gaining ground as a cost-saving tool. "During the Internet boom, companies had so many projects it was tough to turn their attention to a technology that wasn't so spectacular," notes Meisel. "With the slowing of the economy, there's time to look at these technologies, which produce a quick return on investment."

Meisel adds that the companies marketing voice technology have been aided in their efforts by the growing support of warehouse management system (WMS) vendors. "Now it appears that the WMS companies are getting on board," he says. "[WMS] companies see speech as a way of making their product more useful, so they're leveraging it in other sales."

As companies look to boost worker productivity in the warehouse, the voice technology market may be poised for a growth spurt. A recent survey by Logistics Management's sister publication Warehousing Management supports that notion. That survey of almost 400 warehouse managers found that a surprising 18 percent had implemented voice-activated technology in at least one facility and a whopping 73 percent planned to deploy voice technology within one year. For survey respondents, the attraction of voice was its promise of increased productivity and hands-free, eyes-free operations. (See box on this page.)

Though once wary, management is obviously taking a harder look at voice technology's potential to improve a company's distribution operations. "The success of speech in the telephony arena is making it more real to practical, bottom-line minded folks in warehouses and plants," says consultant Markowitz. "This technology is viewed as usable and is marketed as a way of cutting down on labor costs and enhancing efficiency."