magnifying-glass

Why Speaker Independent Voice is the Right Choice for your D.C.

Jennifer Maloney
February 3, 2016

Picking can be the most labor-intensive activity in omni-channel e-commerce order fulfillment operations, especially piece picking operations. Orders requiring rules like a specific package, color, accessory, promotional item, document, or gift wrapping benefit the most from voice technology because it directs and instructs operators on the required shipment and packing rules.

Voice is highly flexible and can change the instructions for the same SKU pick based on a customer, retailer/e-tailer’s specific shipment rule. Voice automatically prompts the picker in a step by step manner with the work rules and integrates a paperless voice instruction step-by-step work process to validate picking. Voice picking outperforms paper, RF terminals, message displays, and many times pick-to-light, especially when accounting for the benefit of its increased accuracy!

The latest speaker independent voice engines have close to 100% recognition rates, while operating in noisy industrial operations, and require zero voice training to operate.

Latest Generation, Lean, Speaker-Independent Voice-Directed Picking

The latest speaker independent voice engines have close to 100% recognition rates, while operating in noisy industrial operations, and require zero voice training to operate.

The latest generation multi-modal voice technology combines a very lean voice command set with integrated hands-free scanning for single touch pick and pack verification.  Integrated scan validation, especially the use of the latest 2-D hands free scanners, significantly increases the speed, flexibility, and accuracy well and beyond traditional speaker dependent voice picking technologies. The latest advancements in speaker-independent voice recognition technology further drive increased productivity and ease of use. The voice engine has close to 100% recognition rates across multiple languages, while operating in noisy industrial operations.  Secondly, zero voice training greatly reduces a worker’s training time to minutes, making it possible to use temporary workers for picking applications.

Speaker-independent natural language voice commands and responses are not only suitable for picking, but support all DC activities such as put to order, replenishment, cycle counting, pallet building, case sorting, and inspection tasks. Voice directs the worker to the location in the most efficient travel path, and directs the operator to perform SKU barcode scanning using a back-of-hand or ring-mounted barcode scanner. It directs and validates each step using a combination of voice commands/responses and hands-free two-dimensional (2D) barcode scanning. 2-D scanning is omni-directional scanning to speed-up by 5% the location, SKU, lot and serial ID barcode capture, and validation steps.

Keeping worker’s hands free and their eyes focused makes an operation safer while allowing operators to hear, see, and verify SKU, quantity, lot, and serial number capture in a fast, accurate, single touch operation. Picking performed with scan validation, vision, and speech is a three-step verification. Multi-verification voice consistently achieves a 99.98% to 99.99% accuracy rate – far higher than voice only technology.

Quite different than first generation voice systems, speaker-independent voice algorithms have continued to advance and now automatically recognize voice responses regardless of an individual’s accent. The system uses a small client message architecture that resides on the small light weight voice-enabled PDA that can operate for 14 hours or more between battery charges.

The Voice Server sends and receives message packets (approximately 25 bytes per message), to transmit the natural voice instructions and receive worker voice and scanning confirmation messages. This uses minimal network bandwidth, and is a fraction of the size of network bandwidth usage compared to many voice picking technologies. The voice server application, database, voice vocabulary, and work logic is server-based. Fulfillment operations can either self-host the solution on their private cloud/virtual server, or host the voice server at a cloud provider site.

The architecture not only handles high-volume picking transactions, but because of its low bytes per message, it minimally impacts the existing facility’s wireless network.  This is different from other voice technologies such as Voice over IP (VoIP), which requires a significant modification to the IT infrastructure network to accommodate the much larger size data packet messages generated by picks.

Pickers are productive in 15 minutes, so even temporary workers can be quickly and easily deployed during peak periods. The voice commands and responses are in a natural voice and support multiple languages. A worker can choose English, Spanish, Polish or many other languages and perform the work in their native language. Speaker-independent voice permits more flexibility in cross-training of workers and interleaving work tasks, leading to better labor utilization for picking, replenishment, and other work tasks.  Additionally, the technology captures and time-stamps all orders and labor usage for performance reporting.

Speaker-independent voice capability highlights:

  • Picking is performed and confirmed following lean efficient order picking processes designed to the order profile.
  • Starts picks within the required zones, selects correct carton, combines voice and scan validation.
  • Directs batch or discrete order picking in the shortest pick path.
  • Picker confirmations are verified by scanning the item barcode, lot and or serial number picked.
  • Directs pick and placement of items to cartons, conveyor, carts, and pallets.
  • Direct work rules and move workers to the “hot” high volume work zones.
  • Labor tracking provides supervisors the ability to analyze individual worker productivity.
  • The process requires no special training. It is remarkably easy to use, learn, and highly accurate, allowing single-touch pick-and-pack validation at near 100% accuracy rates that eliminate the need for secondary order inspection.

Labor tracking provides supervisors the ability to analyze individual worker productivity.

Speaker-independent voice integrates work instructions and verification into each step, for all pick types including pallet, mixed-case, or split-case piece picking. The ability to interleave picking, cycle counting, and replenishment across different order types is readily supported. Voice picking enables work force management to direct and move operators between work zones and work activities.

Look for our next blog: “A Top Tier WCS Increases Productivity in Omni-Channel Distribution.”

Skip to content