支撐向量法在MP3音樂物件分類之應用

Abstract

MP3音樂物件為網際網路上重要的多媒體音樂檔案格式，但是目前對於MP3音樂物件自動化分類的研究卻還不多。在本研究中，我們針對MP3音樂物件的性質進行內涵式分析，然後利用主成份分析以及支撐向量法構造出MP3音樂物件的自動分類器。實驗結果顯示本研究所提出的方法可以適當的分類MP3音樂物件，分類的正確率在訓練集可以達到84.7%，而在測試集可以達到80.3%。In recent years, the MP3 music objects become the popular type of music file in many internet audio application. But, less attention was received to the content-based classification of MP3 music objects.
In this paper, we propose an approach to classify MP3 music objects based on their energy distribution. The technics of PCA (principal component analysis) and support vector machine are used to construct the MP3 classifier. Experiments show that the good performance of an MP3 classification system can be met by the proposed method. The classification correct rate are 84.7% and 80.3% for training data and testing data respectively.