Abstract
The aim of this work is to present machine learning tools for the classification of disease-causing vectors based on audio data. Wingbeat sound recordings of three classes of insects: fruit flies, house flies, and the male Aedes mosquitoes are analyzed using different features. Mel-frequency cepstral coefficients, Gammatone cepstral coefficients, Fourier analysis, and Spectrograms were used as feature detectors. We compare the performance these classification methods and analyze their applicability in wearable devices.