Raga classification based on pitch co-occurrence based features

<p><span>Analysis and classification of raga is the need time especially in music industry. With presence abundance multimedia data on internet, it imperative to develop appropriate tools classify ragas. In this work, an attempt has been made use occurrence pattern pitch based svara (note) for classification. Sequence notes important cue Pitch profile formed. This presents signal along with its statistical distribution can be characterized using co-occurrence matrix. Proposed note matrix summarizes aspect. captures both tonal temporal aspects melody. Ragas differ terms spectral power. K-nearest neighbor (KNN) used as classifier. Publicly available database consisting 300 recordings 30 Hindustani ragas 130 hours audio stored 160 kbps mp3 fileswhich part CompMusic project used. Leave one out validation strategy evaluate performance. Experimental result indicates effectiveness proposed scheme which giving accuracy 93.7%.</span></p>