Please use this identifier to cite or link to this item: https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/3695
Title: Supervised Learning Approach for Classification of Recorded Songs based on Music Structure Similarity
Authors: Peiris, T.S.R
Issue Date: 9-Sep-2016
Abstract: The increase of the music databases on the personal collection and the Internet have brought a great demand for music information retrieval, and especially automatic musical genre classification. Most automatic music genre classification researches have been focusing on combining information from different sources than the musical signal. This thesis presents a novel approach for the automatic music genre classification problem using low-level characteristics of audio signals into high-level hierarchically organized genre taxonomies. The proposed approach uses two feature vectors, Support vector machine classifier with radial basis kernel function. This approach focused mainly on Sri Lankan genres, first to recognize the existence of different genres in Sri Lanka and then to produce an acceptable approach to classify these genres. The short term low-level audio features are derived from 90s audio signal frames with a hop-size of 10ms and all the long term low-level audio features have a frame size of 10 seconds. More specifically, two types of feature vectors for representing frequency domain, temporal domain and cepstral domain short term based audio features and modulation frequency domain long term based audio features are proposed for individual classification. For feature selection purposes, we used wrapper method for short term based feature vector and filtering method for long term based feature vector. The support vector machine classifier with polynomial kernel function is employed as the base classifiers for each individual feature vectors. The best musical genre classification model is obtained from the set of individual results according to their overall accuracy and recall values reported on the dataset from 2 feature sets over four five musical genres respectively. We observed high classification accuracies with respect to both the individual classifiers but the classification model built using temporal, cepstral and timbre features proved to be the best classifier in the end. Finally this approach shows that there exists a need for a genre classification system for local music and that this approach has produced a successful classification model by using different types of domain based audio features. 3
URI: http://hdl.handle.net/123456789/3695
Appears in Collections:SCS Individual Project - Final Thesis (2015)

Files in This Item:
File Description SizeFormat 
2011cs004_Thesis.pdf
  Restricted Access
1.41 MBAdobe PDFView/Open Request a copy


Items in UCSC Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.