Abstract—Respiratory sounds are widely adopted marker of several diseases associated with upper and lower respiratory systems and lungs. Hence, recognition of respiratory sounds is an important step in diagnosis of the several diseases. In this study, it is aimed to recognize normal and asthmatic respiratory sounds. To accomplish this aim, analysis and classification process of the sounds were performed. LPC-based parametric techniques namely Linear Predictive Coefficients (LPC), Linear Prediction Cepstral Coefficients (LPCC) and Weighted Linear Prediction Cepstral Coefficients (WLPCC) techniques were used in analysis and feature extraction process. Linear prediction coefficients, cepstral coefficients and weighted cepstral coefficients were evaluated as characteristic features of the sound signals. In addition, Fuzzy C Means (FCM) clustering algorithm was used to achieve feature reduction. k nearest neighbor (kNN) and fuzzy k nearest neighbor (F-kNN) classifiers were design to classify respiratory sounds as normal and asthmatic sound signals. As a result of this study, the LPC-based parametric techniques were compared in terms of the effect on classification.

Index Terms—linear predictive coefficients, linear prediction cepstral coefficients, weighted linear prediction cepstral coefficients, fuzzy C means, k nearest neighbor, fuzzy k nearest neighbor.

Cite: Fatma Z. Göğüş and Gülay Tezel, "Comparision of LPC Based Parametric Techniques for Respiratory Sounds Recognition," International Journal of Signal Processing Systems, Vol. 6, No. 1, pp. 6-11, March 2018. doi: 10.18178/ijsps.6.1.6-11