HE STRAPS ON A SMALL WIRELESS TRANSMITTER, adjusts a headset, and he’s off. Is he a salesman setting out on a road trip in a state that prohibits hand-held cell phones? Or perhaps a back-up singer for Madonna’s next tour? Or a call center rep? None of the above.

He’s a picker in a warehouse, and thanks to the latest voice and speech recognition technology, he’s off on another hands-free, eyes-on-the-road picking day.

The transmitter at his waist contains information about the orders to be picked, either contained within the unit or beamed to him in real time via wireless communication with the company’s warehouse management system. Through the earpiece of his headset, he hears system-generated speech that tells him which aisle to go to. Once there, he confirms that he is in the right spot — still hands-free — by reading aloud the location’s check-digit code into his headset microphone. If he happens to go to the wrong place, he will hear the system tell him so — and point him in the right direction. If his location code is correct, he will hear the system direct him to the item to be picked. He then confirms what he’s about to pick by reading an item code into his microphone. If he is correct, he picks the item, listens for the next instruction, and takes off for the next location, in a new workday routine that just might revolutionize warehouse operations.

Speech and voice recognition are closely related technologies that, though similar, have a few fundamental differences. Both are programmed to “understand” and generate language, but voice recognition “remembers” an individual’s voice profile, whereas speech recognition recognizes speech patterns and words.

Speech recognition is already commonplace in industries that rely on telephony, such as help desks and catalog call centers. In the warehouse industry, the grocery sector has bought into speech and voice recognition technology, as have some major retailers. But although both vendors and users report that it increases accuracy and productivity and may reduce workplace accidents, other sectors, including the catalog and direct-to-customer businesses, have yet to embrace the technology on a large scale.

“It’s the tip of the iceberg,” says Larry Sweeney, vice president of marketing at Pittsburgh-based Vocollect, one of the three major vendors that serve the warehouse industry.

“We’re talking about an industry that hasn’t had a lot of technological improvements — bar coding in the ’70s, the WMS in the late ’80s, RF scanning in the ’90s,” says Frederick Rost II, director of product marketing and management at Princeton, NJ-based Voxware, Inc. Warehousers “want to see that it works,” he says.

According to Tim Zimmerman, vice president of sales and marketing at Syvox Corp., based in Boulder, CO, the telephony market has helped educate customers about the use of speech technology in commercial applications. “You can track this in the same way as you do traditional hand-held data collection devices. You had one-offs, then standardization, early adopters, and finally widespread use. We see this as following the same track,” Zimmerman says.

Many tracks

At this point, the major vendors say, they have overcome the obstacles that kept the warehouse industry from adopting voice recognition: filtering of background noise and recognition of various languages, accents, and individual voice patterns. With these obstacles addressed, they believe that voice/speech recognition technology is ideal for single-order or small-order picking, as well as other warehouse inventory tracking functions.

“The chain store distribution model has changed,” says Zimmerman. Before, that industry relied on large bulk orders. Now, he says, “there’s the expectation to do single-order and small-order fulfillment [for retail]. Speech recognition enables companies to fulfill those orders.”

It also eliminates many of the impediments to productivity of previous methods. In a paper-based system, a broken pencil can cost a picker valuable time, and reading order specifications from paper while walking can slow down and even endanger the picker on each trip. So can juggling the paper and pencil for picking items that require two hands. With the voice/speech recognition system, there are no pencils or papers to worry about.

By eliminating distractions, says Zimmerman, speech recognition technology “provides an increase in productivity.” It also eliminates the need for manual batch data entry reconciliation. Before voice recognition, says Voxware’s Rost, when the pickers were done, “they’d take their fulfilled pick lists and give them to someone in the shipping department, who’d put it aside for batch data entry.” With the new technology, there are no pick lists turned in by pickers at the end of the shift, so verbal confirmation of the picker’s shift is logged by the system and used for reconciliation. “[The picker] confirms his location, and at the moment that he does that it automatically verifies that he’s at the right place and that he’s picked the right quantity,” says Rost.

Play it safe

There’s also a strong perception that voice and speech recognition technology will make the workplace safer, although none of the vendors or user companies interviewed for this article could quantify it, except anecdotally.

“We have informal information on accident reduction from customers,” says Rost. “One customer had 15 accidents per month. Once they began using our system, they dropped to zero per month.”

Besides picking, voice and speech recognition can be applied to other warehouse functions such as put-away, cross-docking, inventory or cycle counting, flow-through, receiving, and RF loading.

“Companies today are trying to improve their margins,” says Rost. Voice technology “gives them a whole arsenal of things they can do with their business. You can do more with the same number of people. You’re operating more efficiently. A year ago, the question was, ‘Will speech recognition technology work?’ Now it’s ‘Will speak recognition technology work for me?’”

Here’s a rundown of the three vendors’ offerings:

Syvox Corp.

To power its SpeechNet Logistics system, Syvox ( has created an Interactive Speech Interface Specification (ISIS), a universal application that the user can configure to work with any WMS, says Zimmerman. As a speech recognition system, it is speaker-independent. The system is simultaneously multilingual; it can communicate in Spanish and English at the same time and offers a range of other languages, including French, German, Dutch, and Flemish.

The SpeechNet system can communicate directly with the WMS via wireless technology or can store all the data from a shift and download it to the WMS as a cumulative report at the end of the shift. The initial pick order is downloaded to individual pickers’ units, but it also remains on the server in case the picker’s unit is somehow disabled.

Syvox customers typically report 20% to 30% greater productivity over paper-based systems. “We’ve had two customers who actually quote 60% to 100% gains over paper-based systems,” says Zimmerman.

Return on investment typically comes within a year. Companies using the system have realized their ROI with as few as one to two pickers.

Prices hover at less than $4,000 per user for hardware and software. “In the last year, we have seen the per-user cost drop from $6,000,” Zimmerman says. “We anticipate that with the improvements in hardware, we will see that get down to $2,000 to $2,500 per user. It’s becoming more cost-effective for people to deploy, which will increase the number of companies that are able to deploy it, and the ROI will be even faster.” Installation takes between four days and two weeks.


Vocollect’s ( Talkman wireless system can communicate in real time directly with the WMS or through what Sweeney describes as “a piece of middleware that accepts batch-created data from the CMS or WMS and acts as the real-time host computer.”

The system is “speaker-dependent” — that is, it recognizes specific voices by means of the BlueStreak voice recognizer. “Speaker-dependent systems work for 100% of users,” says Sweeney. New users spend 15 minutes talking to the system, so that it will recognize their voice patterns, accents, and languages. They can also record the prompts they will hear in their own voices. The system operates in English, Spanish, French, German, and Dutch, and Sweeney says it can support “virtually any language.”

User companies report an average 75% reduction in errors, says Sweeney, adding, “We’re always within 10% of that on either side. The day you start using voice recognition technology is the day you notice the accuracy improvement.” Productivity typically increases between 10% and 15%, he says.

“The other area where we see improvement is the time it takes to get new employees up to speed,” Sweeney adds. Training operators often takes only half the time it did on the previous system. “We see that as a huge plus for the catalog business; they just about triple the workforce with temporary workers for the holiday season.”

Vocollect recently introduced a new-generation terminal, the Talkman T-2, which, among other features, has a break-away headset connection, so that if the user catches it on anything, it breaks away safely without ripping the headset off the user’s head. The device weighs less than a pound.

Return on investment is typically realized in six to nine months. Vocollect has installed systems for as few as half a dozen users and as many as 220. The price varies, depending on whether Vocollect connects directly to a WMS or the middleware is needed. The Talkman terminal lists at $4,995 per user, with discounts for volume purchases.

Voxware, Inc.

Proprietary voice recognition technology and hardware power Voxware’s ( VoiceLogistics, a wireless Web-based system that allows the warehouse picker to communicate directly with the WMS. The system “adjusts the inventory for the item, it checks that item off the pick list, and it lets the warehouse manager know where this particular worker is in the warehouse and what merchandise is moving in the warehouse,” says Rost.

Voxware’s training and installation staff are versed in warehouse logistics and will help users create new procedures to help them get the most from their investment, Rost says. “We’re going in and automating warehouses that are manual right now. We work with customers to help them rethink what they’re doing, rather than just mimic what they’re doing now.”

VoiceLogistics has been installed in a wide variety of warehouses. The type of operation is more important than the size of the facility, Rost says, noting that the technology is appropriate “where there’s a large number of different products that are handled in a single warehouse.”

The system can handle multiple languages and accents concurrently; systems are installed in English-, French- and Spanish-speaking areas, as well as in Hungary and Germany. VoiceLogistics “enrolls” each user and stores models of the speech that each person uses. Customers report productivity increases of 15% to 40% and up to 73% reduction in errors.

Price depends on the number of users. For 10 to 15 users, for example, the system costs $4,000 to $5,000 per user. In addition, the complexity of the system and whether there is an existing RF infrastructure affect the price.

Karen Berman is a CT-based writer whose work has appeared in a variety of publications, including The New York Times and Wine Enthusiast magazine. She is the author of American Indian Traditions and Ceremonies (World Publications JG Press, 1997).


Tim Beauchamp, senior vice president of operations at Corporate Express, a major business-to-business supplier of office and computer equipment, can’t say enough good things about the voice recognition technology system from Voxware that his company deployed early in 2001. “I’ve been involved in distribution for almost 20 years,” says Beauchamp. “This start-up has been the most bulletproof. And you see improvements quickly. It’s been remarkable.”

The Broomfield, CO-based company was so pleased with the results of its implementation that it will have similar installations in at least half of its 31 distribution sites by the end of 2002. What Corporate Express liked about the Voxware system was that it can be hosted right on the warehouse management system (WMS) unit in each distribution center, with no separate databases. It uses the same communications infrastructure as the radio frequency applications. The company has a central front-end system, which communicates with local warehouse management systems in each of its distribution centers around the U.S.

Beauchamp saw fast improvements in productivity and accuracy, and says return on investment (ROI) comes in less than six months, compared to two years for other major distribution center applications. An unanticipated benefit, he notes, is that workers like the system. “The people are really enjoying it. They like working with the technology.”

Speech Ahoy!

Voice recognition technology has “huge upside potential,” says Larry Warehime, vice president of information systems at Nautica Enterprises, Inc. The New York City-based apparel maker deployed Syvox’s SpeechNet Logistics nearly a year ago in its satellite distribution facility in Virginia. The system has been so successful that the company plans to incorporate it into a new facility now under construction.

“Our picking profile is very broad-spectrum — it ranges from large quantities to individual units,” Warehime explains. Pickers pick merchandise to cartons. A small boutique might receive one carton, while a department stores might receive hundreds. “Speech recognition technology allows us to bind them tightly to the bin and the carton,” says Warehime.

Nautica picked Syvox because of its real-time interface with the apparel merchant’s warehouse management system. “The real-time interface is the key to making the whole thing work and work well,” Warehime says. Another aspect of the Syvox system that appealed to Nautica was that training takes place at the associate level. Warehime points out that “you can have an associate using Syvox within half an hour.”

Easy on the Ice

An ice-cream warehouse presents a very specific set of challenges. Pickers must wear freezer suits and gloves, which make paper picking difficult. Pens often freeze and writing can be near-impossible. In such environments, voice recognition technology is ideal.

Dryer’s/Edy’s (whose product is known as Dryer’s in the western United States and Edy’s in the East) switched to voice recognition technology in April 2000. “Prior to this we were doing everything with paper,” says Brad Adams, Midwest division logistics manager. “We’re a case-pick operation. Pick-to-light is not conducive to a 50-square-foot freezer. We saw voice as the best long-term solution.”

The Oakland, CA-based ice-cream company chose Vocollect because it offered “a more stable system that was simpler to implement.” Dryer’s/Edy’s liked the fact that the Vocollect system downloads an entire pick order into an individual terminal. The terminal sends updates to the main server through wireless RF transmission, but in case the RF is interrupted or the server goes down, the picker keeps working from the list contained in the terminal.

Voice recognition has been a great aid to accuracy, because the company’s average order consists of 75 lines or 75 items. “The more lines per order, the greater benefit the voice technology gives,” says Adams. The company tracks accuracy using perfect orders (not lines); since the implementation of the system, the perfect orders have increased from 93% to 96% (“imperfect” orders were cut in half). The company has also seen a 20% increase in productivity.

Adams also cites the safety benefits of voice recognition technology. “Most pickers will tell you it’s a safer way of working,” he says. “It gives the pickers greater awareness of their surroundings.”

The system’s radio frequency function helps facility managers track picking assignments, locate pickers within the warehouse, and get a handle on the pace and amount of work completed. “Managing the warehouse is easier,” says Adams.

Partner Content

Hincapie Sportswear Finds Omnichannel Success in the Cloud - Netsuite
For more and more companies, a cloud-based unified data solution is the way to make this happen. Custom cycling apparel maker Hincapie Sportswear has leveraged this capability to gain greater visibility into revenue streams, turning opportunities into sales more quickly while gaining overall operating efficiency. Download this ecommerce special report from Multichannel Merchant to more.
The Gift of Wow: Preparing your store for the holiday season - Netsuite
Being prepared for the holiday rush used to mean stocking shelves and making sure your associates were ready for the long hours. But the digital revolution has changed everything, most importantly, customer expectations. Retailers with a physical store presence should be asking themselves—what am I doing to wow the customer?
3 Critical Components to Achieving the Perfect Order - NetSuite
Explore the 3 critical components to delivering the perfect order.
Streamlining Unified Commerce Complexity - NetSuite
Explore how consolidating multiple systems through a cloud-based commerce platform provides a seamless experience for both you, and your customer.