The automatic musical instrument classification taking place in a recording of music has many applications, together with music search through classes, music recommender methods and transcribers. Automatic instrument classification and identification of musical streams has become a difficulty research area over the last few years. In this approach is to classify the audio data based on the instruments. The audio features such as Enhanced Mel Frequency Cepstral Coefficient (EMFCC) and Enhanced Power Normalized Cepstral Coefficients(EPNCC) are used to extract the features for classification of various instrument classes. The classification algorithms such as J48, BFTree, K Star, RandamForest and Bagging are used to classify musical instrument data into classes. Compare with various performance parameters like True Positive Rate, False Positive Rate etc., are used in various classification algorithms. The results shows that the best performance, almost 98% of accuracy, was attained by the classification system using the boosting technique with decision trees.