Indian Musical Instruments Recognition Using LPC and LPI

Abstract: Most Significant Music Industry is today widely supported by Research in the area of Musical information Processing. In the past decade the growth of Digital Music over internet and our day-to-day life has forced researcher to explorethe diverse area of Musical Instrument Identification, Retrieval, Classification and Recognition of large sets of music. In our paper we have explored the ignored area of Indian Musical Instrument Recognition (IMIR) and attempted to build the database for the same. An innovative technique to classify the IMIR using the Modified Linear Predictor Coefficient (LPC) features and Log-Power-Intensity (LPI) is discussed. The proposed method has been tested with experimental results and technique has an accuracy of 96.11% for instrument and 92.45 for individual sound.

Keywords: Indian Musical Instrument Recognition, Linear Predictor Coefficient, Log Power Intensity, Music Instrument Identification, Principle component Analysis, Best First Decision Tree Algorithm