Comparing accuracy of Musical Instrument identification for different features with k-NN and SVM Classifiers

Abstract

Music signal is different from speech signal, therefore for recognizing musical instrument there is a need of designing a different system. Here we have designed an instrument identification system.In this paper we have discussed the role of various features for classifying musical instruments with k-NN and SVM classifier identify musical instrumentsusing Fivemusical instruments: Flute, Piano, Trumpet, Guitar and Xylophone are considered. From each signal of the database we have extracted different features and these for training the classifiers. The trained classifiers tested to identify musical instruments using unseen signals observed that the identification accuracy depends on the features used as well as the classifier. This anal the desired classifier with all features and a desire with a classifier to design the musical instrument identifier system.