McDonalds Voice Order Video Example

By | March 8, 2026
mcdonalds voice order

Last Updated on March 8, 2026 by Craig Allen Keefner

McDonald’s Voice Order Video

Nice video of prototype McDonalds Voice Order. Imagine 3 of these side by side in a NYC McDonalds.  Ambient noise level in a restaurant can easily hit 80 db (Noisy Planet NIH).  This prototype Kiosk ordering has 99% accuracy because a human is in the loop on every order…

If interested in more info – email info@kioskindustry.org

Accents, dialects and languages are TBD but here is response.

It’s a nice demo albeit not a real restaurant with the typical ambient noise.
Correct and the “only reason” why we have 99% accuracy is because we have a human on every call… we have AI which is glorified speech to text to expedite responses, our killer AI application is “Always Indian.” 🙂 
The usual questions regarding accents, dialects, languages along with noise come into play.
Mumble, yell, talk quiet, rev your engine, order from a stadium… it doesn’t matter as long as the human can hear otherwise they’ll ask to repeat or “speak up.” AI cannot do this…
Our automated agent training module has 11 accents where they must take 10,000 orders or have accuracy of 98%+ before going live. 

McDonald’s Kiosk Order from IYC on Vimeo.

More Info

The average noise level in a McDonald’s restaurant can vary, but it is generally quite loud. According to a report by Noisy Planet, noise levels in restaurants can average 80 decibels (dBA) or higher, which is significantly louder than a typical conversation at about 60 dBA1. This level of noise can make it difficult for patrons to have conversations and may even pose a risk to hearing over prolonged exposure. It’s recommended to use earplugs or earmuffs in loud environments to protect your hearing.

From QSR AI

Over the last 5-years, QSR AI has taken restaurant orders with a “Human in the Loop” who presses a button when the AI fails (15% of the time) with approximately 99% accuracy. Less than 1% of our calls are transferred to a live person for order completion.
We’ve challenged ConverseNow, Arria/Ping, and DoorDash to a $5,000,000 accuracy challenge if they surpass our performance. Predictably, they haven’t taken us up on the challenge.
Listen to a “non-staged” Domino’s order here: https://bit.ly/Non_Staged_Dominos_Order
Author: Craig Allen Keefner

Craig Allen Keefner is an industry analyst, content strategist, and longtime authority on self-service kiosks, digital signage, unattended payment systems, and interactive technology. He manages content and industry strategy for Kiosk Industry and The Industry Group, with a focus on kiosk software, hardware-software integration, accessibility, payment compliance, healthcare kiosks, restaurant self-service, and emerging AI automation. Craig has covered the self-service and kiosk industry since the 1990s, tracking how public-facing terminals move from concept to field deployment. His work combines industry research, vendor analysis, operator conversations, standards tracking, trade show coverage, and practical experience with the real-world constraints of kiosk deployments. https://www.linkedin.com/in/kiosk