Musical instrument recognition using k-nearest neighbour and Support Vector Machine

In this paper, we present a model to detect and distinguish individual musical instrument using different feature schemes. The proposed method considers ten musical instruments. The feature extraction scheme consists of temporal, spectral, cepstral and wavelet features. We developed k-nearest neighbor model and support vector machine model to test the performance of system. Our system achieves the 60.43% of recognition rate using k-nearest neighbor classifier with all features. A two prong approach was taken to the multiclass classification which were SVM-one against rest &SVM-one vs. one. The accuracy of SVM in both cases is 73.73% with all features using radial basis function. Using weight factor method knn shows 73% accuracy while SVM shows 90.3% accuracy using exponential kernel function. Using weight factor method knn shows 73% accuracy while SVM shows 90.3% accuracy using exponential kernel function.