文章基本信息

标题：A Comparative Study of Three Algorithms for Efficient Audio Classification
本地全文：下载
作者：Jayita Mitra ; Diganta Saha
期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2015
卷号：3
期号：9
DOI：10.15680/IJIRCCE.2015. 0309099
出版社：S&S Publications
摘要：In this paper, we present a comparative study, on three different efficient feature selection method inclassification of audio files. The main objective is feature selection and extraction. We have selected a set of featuresfor further analysis, which represents the elements in feature vector. By extraction method we can compute a numericalrepresentation that can be used to characterize the audio using the existing toolbox application. In this study Gain Ratio(GR) is used as a feature selection measure. GR has an important role in selection of splitting attribute which willseparate the tuples into different classes. The pulse clarity is considered as a subjective measure and it is used tocalculate the gain of features of audio files. The splitting criterion is employed in the application to identify the class ormusic genre of a specific audio file from testing database. Experimental results indicate that, by using GR theapplication can produce a satisfactory result for music genre classification. After dimensionality reduction best threefeatures have been selected out of various features of audio file and in this technique we get more than 90% successfulclassification result. Principal Component Analysis is used mainly for dimensionality reduction and classify thewestern music database. The encouraging experimental result shows 86.4% successful classification rate using thePCA approach, which indicates a novel way for audio classification. Experimental results In many real worldapplications such as face recognition and mobile robotics, we need to use an adaptive version of feature extractiontechnique. Here, we introduce an adaptive weighted version based on PCA algorithm. Experimental results on westernmusic database demonstrated the effectiveness of the proposed system in audio classification application. TheAdaptively Weighted PCA algorithm discussed here mainly for audio classification which differs from the image basedclassification approaches. Here it is used it to check the efficiency and the performance of classification result of 90.9%using Adaptively Weighted PCA. Through experiments, it is found that, all three algorithms can find a feasible solutionefficiently. But from the data analysis and experimental results, our approach shows a better performance than principalcomponent analysis and AWPCA.
关键词：Data Mining; Feature Extraction; Audio Classification; Gain Ratio; Pulse Clarity; PCA