Analysis of Musical String Instruments using k-NN
Sangeeta Kakarwal, Seema R. Chaudhary · 2020
In musical instrument recognition, audio feature extraction and analysis is one of the important phase before classification. The focus of presented work is on Indian musical string instruments. In this paper, temporal such as Zero Crossing Rate(ZCR) and spectral features such as spectral centroid, spectral rolloff, roughness and brightness, are analysed from monophonic recordings of musical string instruments. String instruments include Guitar, Santoor, Sitar and Sarod. Santoor, Sitar and Sarod are Indian musical string instruments. Classification of musical instruments is done by using k-NN classifier. The system is evaluated for various combinations of features. Using single feature, system has shown 67% best average accuracy for roughness. It is observed that Santoor is having distinct characteristics than other instruments. Sitar and Sarod some how similar in roughness, brightness and spectral rolloff.