Recent advances in voice AI are changing the voice picking experience for everyone: pickers, developers and enterprises. The technology behind voice picking software is mostly invisible; however, it affects end-users directly and significantly. People with prior voice experience that ended up in disappointment due to poor accuracy, high cost, lack of reliability or inflexibility to accommodate changes are rightfully cautious about voice picking software. However, voice technology today is different from what you experienced.

What’s voice picking?

Voice picking, also known as voice-directed picking, pick by voice, or pick-to-voice uses speech recognition and natural language understanding to guide workers to locate, pick, and place items. Voice-directed picking increases productivity by up to 35%, decreases error by up to 90% and shortens the training time from days to minutes. As a result, organizations increase their revenue by providing better service to the customers, reduce labour costs per order, and minimize the cost of shipping a wrong item, resulting in happier customers, employees, and shareholders.

Benefits of voice picking

• Reduced training time: Workers can start after little training, as they’re guided with simple verbal instructions, and are not required to learn complex workflows or memorize checklists. Therefore, warehouses with seasonal workers and high employee turnover save significant time.
• Minimized use of tools and distractions: As voice prompts direct the warehouse operators, they do not need to pick up, read, and put down instructions at work. They focus on main tasks: visually locate correct items, pick the right quantity, and place them. Tasks such as data entry or following written instructions result in picking the wrong item or quantity and wasting time between bins. Simplifying the process and limiting distractions increase speed and accuracy.
• Improved workplace safety: Leaving pickers’ hands and eyes free, especially while dealing with heavy objects, or using sharp items improves workplace safety significantly.

Unsurprisingly, large distribution centres and warehouse management system providers have joined the voice trend. The voice-directed warehousing solution market is expected to grow from $1.4 billion in 2020 to $4.8 billion in 2031. There will be many more enterprises adopting voice-directed systems in this decade. If you consider joining the voice revolution whether as a supplier or buyer, there are 3 major things that you should know to evaluate alternatives: Connectivity, Ease of Use, and Cost.

Things to know

• Connectivity: For consistent and guaranteed real-time experience voice data should be processed fully on-device with no connectivity dependency to perform tasks. Otherwise, voice-picking software which relies on the cloud for voice data processing will hinder the experience and productivity. Even for simple tasks such as understanding where to go next, when a picker asks a question, the data will be transmitted to the cloud, processed on the cloud, and transmitted back. Any connectivity issues in the warehouse, with the internet, or with the cloud service provider hinder the picker’s productivity.

Source: picovoice.ai

• Ease of Use: Ease of use should be evaluated for both pickers and developers.

1. Ease of Use for Pickers: Demographics of the warehouse workers vary: Different age groups, educational backgrounds, dialects and accents… Finding a solution that works across different accents and in noisy warehouse environments with high accuracy and that requires no or limited technology literacy is key.
2. Ease of Use for Developers: With the advances in technology, training voice models became as easy as downloading anything from the internet. Any voice vendor that requires you to train the algorithm with your voice files will limit your agility and return poor accuracy across various accents. Another red flag is long-term contracts without providing updates and upgrades, as their attachment to end-user success will be low. Although voice recognition technology has achieved a certain level of accuracy and it gets harder to improve accuracy, even a 1% improvement may mean a 33% decrease in error rates.

• Cost: Every IT decision-maker knows that the cost of a product is different from the total cost of ownership. Two things may skyrocket your total cost of ownership significantly: 1) Development Cost 2) Cloud Bill

1. Development Cost: If developing a prototype takes months, then it means updating the software will take longer as well. The developer interface should be easy to accommodate development and changes in the business (new products, aisles) without affecting the operations. The longer time required for time-to-market means allocating more developers and postponing capitalizing on benefits, resulting in higher costs for both suppliers and users of the voice picking software.
2. Cloud Bill: Cloud providers charge based on usage: number of API calls, minutes of processed voice data. While developing a prototype it might not be a problem. However, when workers start picking millions of items, the bill will go up. Major cloud providers such as Google charge $0.06 per minute of usage. Assuming that you have a warehouse with 1000 employees working for a 7-hour shift per day, your monthly cost will be close to $750,000. With on-device voice processing computing, the cost could go down up to two orders of magnitude.

Every second spent between bins costs enterprises money. Large enterprises with access to massive resources have already started modernizing their warehouses, resulting in revenue and cost advantages over their competitors. There’s enough information to showcase the advantages of voice picking, now the question is to find the right partner to offer the best experience for your employees and cost-benefits to your company to join the voice revolution in warehouses!

About the author

Dilek Karasoy is Marketing Director at Picovoice, the end-to-end voice AI platform. She holds an MBA from the Rotman School, University of Toronto and is passionate about diversity in the tech industry. Prior joining to Picovoice, Dilek worked at Deloitte Consulting, Intel and IBM.