Selecting Proper Features and Classifiers for Accurate Identification of Musical Instruments

Abstract

Selection of effective feature set and proper classifier is a challenging task in problems where machine learning techniques are used. In automatic identification of musical instruments also it is very crucial to find the right set of features and accurate classifier. In this paper, the role of various features with different classifiers on automatic identification of musical instruments is discussed. Piano, acoustic guitar, xylophone and violin are identified using various features and classifiers. Spectral features like spectral centroid, spectral slope, spectral spread, spectral kurtosis, spectral skewness and spectral roll-off are used along with autocorrelation coefficients and Mel Frequency Cepstral Coefficients (MFCC) for this purpose. The dependence of instrument identification accuracy on these features is studied for different classifiers. Decision trees, k nearest neighbour classifier, multilayer perceptron, Sequential Minimal Optimization Algorithm (SMO) and multi class classifier (metaclassifier) are used. It is observed that accuracy can be improved by proper selection of these features and classifier.