Like many businesses competing in a global market, Sony Corporation is challenged to do more with less.
The electronics company provides leading-edge products like Blu-ray Disc players, Bravia televisions, Cyber-shot digital cameras and Handycam Camcorders to consumers across the globe. In recent years, its wholly-owned Canadian subsidiary, Sony of Canada Ltd., experienced a rapidly changing business model of increasing online orders — particularly an increase in smaller orders — coupled with growth.
Every day, Sony factories around the world ship approximately 1,500 products to Sony of Canada’s distribution centres in Coquitlam, B.C., and Whitby, Ont., which in turn collectively process 18,000 orders per month. Sony of Canada sometimes had trouble keeping up.
The company’s growth quickly exposed the weaknesses in the paper-based and RF systems it used for distribution. Paper was cumbersome and it provided no locator system to help employees find products quickly and easily. While RF scanning was quicker and faster than paper, picking with RF consisted of too many steps, which increased the opportunity for errors. Workers had to input into the handheld unit, view the screen, put down and pick up the device, and the system was often slow to respond.
Sony of Canada was looking for a way to improve its efficiency and effectiveness, service levels and costs, so that it could remain competitive in a challenging global market.
At the recommendation of RedPrairie, Sony of Canada’s WMS provider and technology partner, the company decided to investigate the potential of Vocollect Voice, voice-enabling technology used in manufacturing plants and distribution centres.
“I was introduced to the voice technology at RedPrairie’s user conference and, of course, immediately I’m thinking, ‘Wow, this is amazing stuff. We could certainly use it,’” recalls Rick Courtin, business process manager, Supply Chain Group, Sony of Canada.
Based on observing voice in action at another Vocollect customer site and consultation from RedPrairie, the company could see that the integration of Vocollect Voice into an upcoming WMS upgrade would be an excellent way to achieve maximum payback in process improvements across its distribution operations.
Handling bulky RF units, looking at a screen for instructions and inputting data would no longer slow down their distribution process. With voice, they could move to a hands-free/eyes-free environment.
“With voice, the beauty of it is that it’s a very fluid capability. You don’t have to put things down and then go back to them. So you’re just basically talking and you’re walking and you’re doing the job as you go,” explains Courtin.
How it works
“Vocollect and RedPrairie have been long-time partners, and we work very closely together so that we can build a real-time direct interface between the RedPrairie WMS business logic and the process logic that exists inside the Vocollect system,” explains Gary Glessner, vice-president, Sales-Americas, Vocollect. “We do this because we want to make it easy for folks like Rick, Sony and our other customers to seamlessly be able to deploy voice with minimal extra effort of any kind.”
When an order is sent to the WMS, Vocollect Voice translates it into voice commands, and the system talks to the worker through a headset with a microphone.
“So instead of an operator having to, with an RF device, scan or look at the screen and enter key pushes to tell the computer to advance to the next step, all of that is just being done by speech commands,” explains Glessner.
The system includes a text-to-speech engine, available in many different languages, which takes the data that the WMS system is sending out and automatically converts it to speech commands. The worker then responds into the microphone. Voice recognition algorithm software resides inside the voice client that runs on the mobile computing device, and it does the interpretation of what the worker is saying and turns that into data that goes back to the WMS system.
The recognition algorithm that Vocollect uses is speaker dependent. Each worker will spend 20 minutes or so when they first start using the voice system to train their individual voice template, which is stored on a small computer or server. Every time the worker logs in and starts a shift, their specific voice template is downloaded to their specific mobile computer.
“When that worker trains the template, they can speak in whatever accent, dialect [or] language,” says Glessner. “As long as they’re consistent, the system will consistently recognize them.”
Workers using the system can also choose their own settings. They can select whether the voice is male or female, and adjust the pitch, volume and speed of the voice.
Vocollect at Sony of Canada
Sony of Canada uses a consolidated picking methodology — cluster picking — for processing less than master carton quantities. Initially Vocollect Voice was piloted for cluster picking for parcel shipments because of its process fit and the high number of picks — about 65 percent of their transactions go through the parcel mode. After a short training process (as little as one hour), workers were up and running.
Today, orders are batched, and a bulk pick of product is then brought forward to a staging area where the individual orders are picked, packed and shipped through the use of Vocollect.
Since that time, the company has expanded voice to the cycle counting workflow, which is the process of counting inventory. Sony of Canada performs cycle counting with voice every day. Cycle counts are generated and operations staff are sent instructions via the WMS to the voice system to perform the counts. Location and SKUs are scanned and counted as blind counts. If the count does not agree with the system, the operator is asked to verify. If the second verification doesn’t agree with the system count, an audit count — performed by supervisory or management staff — is automatically generated. Using Vocollect for cycle counting has resulted in increases in accuracy through hands-free counts, and it has eliminated costly and time-consuming periodic audits for Sony of Canada.
Voice picking not only requires fewer steps than using an RF device, but workers are able to perform each of those steps more rapidly and productively than if they had to hold and put down an RF device to perform their work, look at a screen for instructions and use the device to scan or key input information.
“So that’s where you pick up all the productivity gains with voice,” says Glessner. “[It] is not only reducing the number of steps, but being able to do those steps much more productively and efficiently.”
The ability to accurately and quickly fulfill orders, to manage seasonal volume hikes, to help the company sustain a thriving Internet business with large numbers of smaller orders, and to do more with less are all benefits the voice solution has brought.
“With our old paper-based system and its inherent inefficiencies, physical inventory took us four business days. Moving to RF reduced the inventory process to just a day and a half. But with Vocollect Voice, now we spend only half a day completing physical inventories,” says Courtin.
Since implementing Vocollect Voice, Sony of Canada has realized improvements of 30 percent in throughput and 15 percent in accuracy, as well as improvements in safety, with the hands-free/eyes-free feature of the voice system.
Reduced training time is also a big benefit, Courtin says. It only takes an hour or so for technical training on the voice system, and about a week for the full operational training. In the past, it would take employees two to three months to complete RF scanning and operational training.
These improved efficiencies have allowed the company to reduce its temporary headcount by five workers.
“We’ve often had to bring in temporary help to augment the full-time staff, and we’ve basically cut right back,” says Courtin. “The efficiency gain has been such that we didn’t have to hire nearly as many temps as in the past.”
Voice technology has also allowed Sony of Canada to better manage the seasonality of the business, with the peak order time being September through December. The voice system provides workers with the ability to nimbly move from one assignment to the next with no downtime, even during peak periods.
Voices in the plant
Sony of Canada has seen such huge benefits from using voice for cluster picking and cycle counting, that it plans to expand Vocollect Voice to full pallet picks, put-away, and Less than Trailer Load (LTL) picking. (They are currently in the middle of piloting it in LTL picking.) However, applications span beyond distribution.
“What we see is that companies are clearly seeing, especially in manufacturing applications, that the hands-free/eyes-free accuracy and productivity — that optimal combination of accuracy and productivity — is allowing companies to deploy voice in applications such as the feeding of assembly lines, the feeding of manufacturing cells [and] kitting applications,” says Glessner, adding that even inspection was an early application for Vocollect Voice. “It’s predominantly in use today in distribution centre operations, but…many manufacturers are discovering that in just-in-time assembly cells and manufacturing lines, voice has a significant value-add.”
“There are customers that have a belief that voice adds complexity to an implementation,” says Scott Dunnington, service director with RedPrairie Corporation. “What we really find is that voice reduces the complexity. It makes for more successful projects. It makes for happier customers, happier users and better all around project results.”
And Sony of Canada is proof. After using Vocollect Voice for four years, and seeing the huge benefits — including an ROI of less than one year — Courtin agrees.
“I can tell you from the staff that use it that it’s very much a pleasure for them to work with that technology.”
It makes their jobs easier, and they are more productive, efficient and effective — the exact combination that Sony of Canada needs to remain competitive and ensure that its products arrive to customers on time, every time.
This article originally appeared in the May 2012 issue of Manufacturing AUTOMATION.