Voice ordering at the self-service kiosk? Coming right up

Editor's Note: This is part two in a two-part series. Part one is available here.

Guests will be ordering their meals faster than ever, thanks to self-service technology. And they will not feel like an anonymous number in the process.

That's because the self-order experience will be verbal, as opposed to using a touchscreen, and the guest will be able to customize their order in their own words and have that order recognized in real time.

Such is the future envisioned by Ben Kaplan, owner of the Boston-based PLNT Burger fast casual chain, and Ken Sutton, CEO of Yobe Inc., a developer of AI-enabled voice solutions. Kaplan and Sutton are working on a voice solution that will allow guests to speak their order at a kiosk and pick it up in a matter of minutes.

"You didn't pull out a wallet, you didn't swipe a card," Sutton told this website in a phone interview, describing the voice-activated self-service transaction. "Everything is done with voice."

The partners plan to introduce it to PLNT Burger's freestanding New York stores in the near future before expanding it to all 14 stores.

A new era unfolds

As noted in part one of this two-part series, Kaplan, a long-time student of restaurant technology, believes it is now possible to provide a customized experience for guests that recognizes them by their individual voice, pulls up preferences and responds based on emotion, intent and mood.

"Restaurants have been inundated with technology solutions focused on making the guest experience frictionless," Kaplan said in a prepared statement. "While those systems have done a good job with providing efficient service, they have missed the mark on providing genuine hospitality to our guests. With Yobe, we can now use voice to create a more authentic connection and still get the efficiency benefits that our guests have come to expect from ordering tech."

The partners are also working with SoundHound, a provider of an AI-enabled voice solution that can understand natural human speech and integrate the voice order with the restaurant's existing order process.

'Hands-free texting'

During the recent CES show in Las Vegas, Yobe demonstrated its voice solution to visitors on the crowded, noisy trade show floor. During the demos, Sutton referenced Yobe ordering as "hands-free texting."

A digital screen at the exhibit displayed users' words in text form as they spoke into a phone.

The text immediately appeared on the screen at the sound of the biometrically linked user's voice. When another person standing next to the biometrically linked user began speaking loudly, that non-linked person's words did not appear on the screen. The software filtered out the unneeded sound.

"I can hear them over the noise, but more importantly, I know if they were serious or if they were joking," Sutton said. "All of that is tracked in the voice signal."

Sutton launched Yobe in Boston in 2016 along with Hamid Nawab, who holds a doctorate in signal processing and machine perception. The duo developed a voice information extraction and speaker tracking solution to extract voice data — biological, linguistic and acoustic data — in noisy spaces that typically compromise the accuracy and effectiveness of voice interfaces.

The solution does not require cloud-based computing or an Internet connection and can operate on various hardware platforms to filter out background noise and accurately identify and interpret the user.

Restaurant voice tech evolves

Voice ordering is hardly new to the restaurant space. Consumers have used voice assistants to place phone orders for years.

Drive-thru is the second arena in which restaurants have leveraged voice orders, with a number of QSRs offering voice-to-text ordering.

Self-order kiosks, however, mark a fairly new frontier.

During a recent webinar, Ben Bellettini, vice president of sales for restaurants at SoundHound, said more retailers are asking about the possibility of integrating voice ordering with self-serve kiosks. He noted that "sound isolation" is an important consideration for self-order kiosks. It is important for the system to identify the audio for the individual placing the order.

For PLNT Burger, SoundHound will transcribe the voice order into the restaurant's existing order process and any necessary back-of-the-house operations.

"Their (Yobe's) software will run on the local device and ensure that we receive clear audio," Mike Lauricella, head of channel partnerships at SoundHound, said in an email. "Yobe technology complements SoundHound's voice AI to enable seamless functionality, even in noisy and distracting environments. We want to give restaurant customers the confidence to speak naturally and still be heard and understood."

In search of clear audio

Sutton agreed soundly, no pun intended.

"If you're trying to use any kind of back-end automatic speech recognition or speech-to-text platform, you can't do anything in the voice space without clean speech," Sutton said.

At the time of this report, Sutton and Kaplan were still finalizing the hardware to host the PLNT Burger voice solution. The kiosk must support more than one microphone.

In the future, the guest will receive an option at the end of their first order to save their voice data for future orders. The data sits on the restaurant's server, not in the cloud.

"Your data never leaves the server that it's on," Sutton said.

"We see voice as being an interesting revenue driver, not just a nice-to-have because it creates a better UI," he said.

"If you can remove the barrier of entry, which is, 'Can I get clean data from voice in these noisy environments' … the ability to be innovative around voice technology and the personalization that comes with that … is now a revenue producing type of conversation."