Pre-Processing and Classification of Cough Sounds in Noisy Environment using SVM

This paper presents a methodology for automated analysis of cough sound signals using Support Vector Machines (SVM). Pre-Processing is necessary for suppression of noise from cough signals; and is the prime concern in accurate prediction of various respiratory disorders. Herein, a 'Fourth Order Butterworth High Pass Filter (FOBHPF) is therefore applied at the pre-processing stage followed by features extraction and classification using SVM. The performance evaluation during the simulations has carried out using Signal-to-Noise Ratio (SNR) parameter at output of the pre-processing stage and accuracy parameter at the end of classification stage.