Basic Study on Taste Identification Using Audio Input/Output

This research aims to develop a new taste sensor technology utilizing the audio input/output of common devices. We propose a low-cost "taste sensor" that measures the electrical properties of liquids via the audio I/O terminals of PCs and smartphones. This sensor identifies taste from the transfer function of liquid samples measured by swept sine wave signals. Using salt water, citric acid, sugar water solutions at different concentrations as samples, we apply a swept sine to the samples via the audio output and measure the response from the audio input. Convolving this response with an inverse filter gives the impulse response, whose fast Fourier transform yields the amplitude and phase frequency characteristics. Principal component analysis and random forest classification methods then identify tastes from this data, achieving 93% average accuracy across samples. This technology enables consumers to easily digitize taste information by utilizing readily available audio hardware, paving the way for new taste sensor development.