臺灣博碩士論文加值系統

隨著現今數位音樂資料庫在網路上越來越普及，對管理者而言如何有效地管理這些數量龐大的音樂資料庫，以及對使用者而言如何快速地從資料庫中查詢到想要之歌曲，事先針對資料庫中每首歌曲根據不同的音樂風格作分類可以解決上述的問題；本篇論文將提供一個以歌曲內容為基礎之自動化音樂風格分類系統來幫助管理一個龐大的音樂資料庫，因此兩個新穎的音樂特徵，低頻能量比率 (low frequency energy ratio - LFER) 以及能域信號編碼 (energy domain signal coding - EDSC)，將提出來做音樂風格分類。低頻能量比率擷取出特定音樂型態之低頻能量當作特徵，能域信號編碼研究能量或音量變化來估測歌曲之節奏資訊。在實驗結果部份除了採用我們所提出的兩特徵外，將結合現有的特徵梅爾倒頻譜係數 (Mel frequency cepstral coefficients - MFCCs) 以及以八度音為基礎的頻譜相對係數 (octave-based spectral contrast - OSC) 來提高分類正確率。除此之外，線性區分分析 (linear discriminant analysis) 亦加入我們系統更進一步地改善整體分類正確率以及降低特徵維度。

Recently, with the construction of digital music libraries, it is important to efficiently manage a large music database. It will be helpful to provide a content-based music genre classification system for managing a large music database. Therefore, two novel music features, low-frequency energy ratio (LFER) and energy domain signal coding (EDSC), will be proposed for music genre classification. LFER extracts the energy of low-frequency components as the characteristics for a specific music type. EDSC aims to exploit the variations of energy or loudness to estimate the rhythmic information of a music track. Experimental results have shown that when the proposed LFER and EDSC features are incorporated with existing features Mel-frequency cepstral coefficients (MFCCs) and octave-based spectral contrast (OSC) the classification accuracy will be improved. In addition, by using linear discriminant analysis (LDA) the classification accuracy can be further improved whereas the feature dimension is reduced.

CHAPTER 1 INTRODUCTION
1.1 Motivation
1.2 An Overview of Music Genre Classification Systems
1.2.1 Feature Extraction
1.2.2 Feature Selection
1.2.3 Feature Classifier
1.3 Outline of the Thesis
CHAPTER 2 THE PROPOSED MUSIC GENRE CLASSIFICATION METHOD
2.1 Feature Extraction
2.1.1 Mel-Frequency Cepstral Coefficients (MFCCs)
2.1.2 Octave-based Spectral Contrast (OSC)
2.1.3 Low-Frequency Energy Ratio (LFER)
2.1.4 Energy Domain Signal Coding (EDSC)
2.1.5 Feature Vector Normalization
2.2 Linear Discriminant Analysis (LDA)
2.3 Music Genre Classification Phase
CHAPTER 3 EXPERIMENTAL RESULTS
3.1 Experiment 1 – Four Music Genres Classification
3.2 Experiment 2 – Seven Music Genres Classification
3.2.1 Direct Classification
3.2.2 Hierarchical Classification
CHAPTER 4 CONCLUSIONS
REFERENCES

[1]G. Tzanetakis and P. Cook, “Musical Genre Classification of Audio Signals”, IEEE Transactions on Speech and Audio Processing, Vol. 10, No. 3, pp. 293-302, July 2002.
[2]G. Tzanetakis, G. Essl, and P. Cook, “Automatic musical genre classification of audio signals”, in Proc. of ISMIR, 2001.
[3]Heather D. Jennings, Plamen Ch. Ivanov, Allen de M. Martins, P.C. da Silva, and G.M. Viswanathan, “Variance fluctuations in nonstationary time series: a comparative study of music genres”, Physica A, Vol. 336, pp. 585-594, 2004.
[4]M.Bigerelle and A. Iost, “Fractal dimension and classification of music”, Chaos, Solitons and Fractals, Vol. 11, pp. 2179-2192, 2000.
[5]T Li, M. Ogihara, and Q. Li, “A Comparative Study on Content-Based Music Genre Classification”, in Proc. of Annual ACM Conf. on Research and Development in Information Retrieval, pp. 282-289, July/August, 2003.
[6]K. West and S. Cox, “Features and classifiers for the automatic classification of musical audio signals”, in Proc. of International Symposium on Music Information Retrieval (ISMIR), 2004.
[7]D. N. Jiang, L. Lu, H. J. Zhang, J. H. Tao, and L. H. Cai, “Music type classification by spectral contrast feature”, in Proc. of IEEE Int. Conf., Vol. 1, pp. 113-116, 2002.
[8]K. Umapathy, S. Krishnan, and S. Jimaa, “Multigroup classification of audio signals using Time-Frequency parameters”, IEEE Transactions on Multimedia, Vol. 7, No. 2, pp. 308-315, April 2005.
[9]M. F. McKinney and J. Breebaart, “Features for audio and music classification”, in Proc. of the 4th Int. Conf. on Music Information Retrieval, 2003.
[10]J. Jose Burred and A. Lerch, “A hierarchical approach to automatic musical genre classify-cation”, in Proc. of the 6th Int. Conf. on Digital Audio Effects, pp. 8-11, September 2003.
[11]T. Li and M. Ogihara, “Music Genre Classification with Taxonomy”, in Proc. of IEEE Int. Conf. on Acoustics, Speech, and Signal Processing, Vol. 5, pp. 197-200, March 2005.
[12]J.-J. Aucouturier and F. Pachet, “Representing musical genre: A state of the art”, Journal of new musical research, Vol. 32, No. 1, pp. 83-93, 2003.
[13]M. E. P. Davies and M. D. Plumbley, “Beat tracking with a two state model”, in Proc. Int. Conf. on Acoustic, Speech, and Signal Processing (ICASSP), 2005.
[14]W. A. Sethares, R. D. Robin, and J. C. Sethares, “Beat tracking of musical performance using low-level audio feature”, IEEE Transactions on speech and audio processing, Vol. 13, No. 12, March 2005.
[15]G. Tzanetakis, A. Ermolinskyi, and P. Cook, “Pitch Histogram in Audio and Symbolic Music Information Retrieval”, in Proc. IRCAM, 2002.
[16]T. Tolonen and M. Karjalainen, “A Computationally Efficient Multipitch Analysis Model”, IEEE Transactions on Speech and Audio Processing, Vol. 8, No. 6, pp. 708-716, November 2000.
[17]R. Meddis and L. O’Mard, “A unitary model of pitch perception”, Acoustical Society of America, Vol. 102, No. 3, pp. 1811-1820, September 1997.
[18]C. Xu, N. C. Maddage, and X. Shao, “Automatic music classification and summarization”, IEEE Transactions on Speech and Audio Processing, Vol. 13, No. 3, pp. 441-450, May 2005.
[19]S. Katagiri and C. H. Lee, “A New Hybird Algorithm for Speech Recognition Based on HMM Segmentation and Learning Vector Quantization,” IEEE Trans. on Speech and Audio Processing, Vol. 1, No. 4, pp. 421-430, October 1993.
[20]R. Duda, P. Hart, and D. Stork, Pattern Classification, New York:Wiley, 2000.
[21]L. Breiman, J. H. Friedman, R. A. Olshen, and C. J. Stone, “Classification and regression trees”, Wadsworth and Brooks/Cole Advanced books and Software, 1984.
[22]M. Grimaldi, P. Cunningham, and A. Kokaram, “An evaluation of alternative feature selection strategies and ensemble techniques for classifying music”, in Proc. 14th European Conf. on Machine Learning (ECML), 2003.
[23]R. Vergin, D. O’Shaughnessy, and V. Gupta, “Compensated Mel Frequency Cepstrum Coefficients”, in Proc. of Int. Conf. on Acoustics, Speech, and Signal Processing, May 1996.
[24]E. D. Chesmore, “Application of time domain signal coding and artificial neural networks to passive acoustical identification of animals”, Applied Acoustics, Vol. 62, pp. 1359-1374, 2001.