Prototype of Food Hardness Measurement App Using Swept-Sine Technique

The Swept-Sine technique using a smartphone makes available so that consumers can check food quality by themselves. Sweep sounds apply to gelatine samples with various hardness from the smartphone speaker, and response sounds are captured with a microphone of the identical smartphone. Impulse response is calculated by convolution integral between the sweep sounds and response sounds. Between frequency distribution of the impulse response and the hardness of the samples are investigated. Measurement using k-fold cross- validation methods and support vector machine shows an average accuracy of 0.958 in hardness classification. As a result of 72-time measurement, principal component analysis showed clear separations among hardness groups.