Related Articles
You take a bite of a carrot. Snap! You nosh on soup. Slurp. You bite into a cookie. Crunch. The foods you eat make different sounds as you mash them with your teeth. University at Buffalo computer scientist Wenyao Xu decided to test whether technology can track food intake based on noises from eating.
Xu developed a necklace called AutoDietary. It uses a small microphone to listen to the sounds of chewing. Those sounds are matched up to an audio library through a smartphone app. “Each food, as it’s chewed, has its own voice,” Xu said in a statement released by the university Wednesday.
Xu conducted a small study using 12 participants. They were each given water, as well as apples, carrots, cookies, peanuts, walnuts and potato chips. The necklace was able to determine the right food 85 percent of the time.
Related stories
- Google cooking up AI to count calories in your food photos
- Hit every fitness goal with these apps
- Calorie-counting microwave on the horizon?
The resulting study, titled “AutoDietary: A Wearable Acoustic Sensor System for Food Intake Recognition in Daily Life,” was published in the February issue of the IEEE Sensors Journal.
Tracking food intake can be challenging. Xu hopes a necklace like AutoDietary could one day help people with dietary issues, obesity or diabetes track and manage what they eat.
AutoDietary is still in the early stages of development. While it can accurately hear some foods, it has trouble telling the difference between comestibles that may sound the same when eaten. For example, it wouldn’t be able to distinguish between a plain pancake or a pancake soaked in sugary syrup.
Xu’s future solution to AutoDietary’s shortcomings would be a biomonitor for tracking blood sugar levels and other biological measurements to help determine the nutritional value and makeup of food. Combine that with the necklace’s intelligent hearing capabilities and you could keep a handle on your eating habits using a wearable as your food coach.
Now, to make it look more fashionable…
