Measuring crunchiness in food

Measuring the level of "crunchiness" in food and snacks is not standardized. Food manufacturers use a combination of texture analyzers and consumer panel testing to measure crunchiness in their food products. In this project, we aim to develop a standardized crunchiness measuring system that records the audio signals from the food when it is cracked, broken or crushed. Using signal processing, machine learning and AI models, the resulting audio file is analyzed to quantify the crunchiness perception of a human. Experiments were performed with multiple types of snack items such as chips, fries, cookies, chocolates etc., to find the ideal setup to break different types of food items such that the machine learning and AI models can accurately learn to differentiate between different levels of crunchy food. This project is in collaboration with Hogeschool VIVES and many industrial partners such as Kellanova, Lotus Bakeries, Choprabisco, Vandemoortele and many more. This is an ongoing project, where I develop data collection protocols for optimizing the setup and use machine learning and AI to produce models with high predictive performance.