臺灣博碩士論文加值系統

吉他效果器大量運用在電吉他的彈奏上，任何彈奏電吉他的樂手都會研究效果器以找出屬於自己個性的效果，電吉他的樂手在彈奏他人的作品時也常希望能模仿彈奏者的音色。吉他有許多種不同的效果器，每種效果器也有許多不同的參數需要調整，因此要模仿彈奏者的音色需要長時間調整的一件事。本論文使用Hammerstein-Wiener 的模型以及類神經網路(Neural Network)來模仿彈奏者破音(Distortion)的音色，並針對延遲效果利用了自相關係數(Autocorrelation)方法來自動偵測延遲效果器的參數並進而模仿音色，在延遲以及破音二種效果同時使用的情況，結合Hammerstein-Wiener 以及自相關係數的方法來模仿音色。對於破音的模仿可達到99%的相似度，偵測延遲效果器的三個參數之平均正確率為90.27%，延遲以及破音同時使用的情況，其模仿平均可達95%的相似度。

Guitar effects are commonly used in playing electric guitar. The musicians of electric guitar player will study how to find out own guitar tone. When they playing others’ music, guitar players want to imitate the original guitar tone. Because there are a lot of different guitar effector, and there are many parameters on each effector, so the original tone imitation is very tough and need to take much time for learning. In this thesis we use computer to simulate distortion effect based on Hammerstein-Wiener and neural network. For delay effect simulation, autocorrelation method can be applied in delay parameters estimation, then we can use the information of parameters to simulate delay effect. For the delay and distortion effect, we use Hammerstein-Wiener and combine the autocorrelation to simulate delay and distortion situation. The fit of distortion simulation with neural network is 99%, the accuracy of three delay parameters with autocorrelation method is 90.27%, and the fit of delay and distortion simulation with Hammerstein-Wiener and autocorrelation is 95%.

口試委員會審定書 #
誌謝 i
中文摘要 ii
ABSTRACT iii
CONTENTS iv
LIST OF FIGURES vi
LIST OF TABLES viii
Chapter 1 Introduction 1
Chapter 2 Background 4
2.1 Effector of Electric Guitar 4
2.1.1 Electric Guitar Effects 4
2.1.2 Distortion Effect 6
2.1.3 Delay Effect. 9
2.2 Hammerstein-Wiener System Identification 11
2.3 Autocorrelation for Delay Parameter Detection 12
2.4 Neural Network 14
Chapter 3 Distortion Architecture and Delay Parameters Extraction 18
3.1 System Architecture 18
3.2 Distortion Architecture Construction 23
3.3 Delay Parameter Extraction and Simulation 25
3.3.1 Neural Network for Delay Parameters Extraction 25
3.3.2 The Method of Autocorrelation for Delay Parameters Extraction 26
3.4 Distortion and Delay Effects Simulation 28
3.4.1 Neural Network and Hammerstein-Wiener for Distortion and Delay Simulation 28
3.4.2 Hammerstein-Wiener and Autocorrelation for Delay and Distortion Simulation 29
Chapter 4 Experiments and Results 31
4.1 Experiment with the Distortion Simulation 31
4.1.1 Experiment with the Distortion Simulation for Hammerstein-Wiener System Identification 31
4.1.2 Experiment with the Distortion Simulation for Neural Network 33
4.2 Experiment with the Delay Parameter Extraction 35
4.2.1 Experiment with the Neural Network for Delay Parameter Extraction 35
4.2.2 Experiment with the Autocorrelation for Delay Parameter Extraction and Simulation 36
4.3 Experiment with the Delay and Distortion simulation 38
4.3.1 Experiment with the Distortion and Delay Simulation for Neural Network and Hammerstein-Wiener 38
4.3.2 Experiment with the Distortion and Delay Simulation for Autocorrelation and Hammerstein-Wiener 42
Chapter 5 Conclusions 45
REFERENCE 46
Appendix 47

[1]K. Shi, X. Ma, and G. Tong Zhou, “Adaptive Acoustic Echo Cancellation in The Presence of Multiple Nonlinearities,” IEEE International Conference on Acoustics, Speech and Signal Processing, 2008
[2]T. Li, “Musical genre classification of audio signals,” IEEE Transactions on Speech and Audio Processing, vol. 10, no. 5, pp. 293-302, 2002.
[3]C. Silla Jr, et al., “A machine learning approach to automatic music genre classification,” RIAO, Pittsburgh, 2007
[4]S. Doraisamy, et al., “A study on feature selection and classification techniques for automatic genre classification of traditional malay music,” in Proc. ISMIR 2008, pp. 331-336
[5]M. Marolt, “Transcription of polyphonic piano music with neural networks,” Electrotechnical Conference, vol. 2, pp. 512-515, 2000
[6]蕭力維(國立國立臺灣大學), “吉他效果器效果辨認與延遲估計” 國 立 臺 灣 大 學 電機資訊學院 電機工程學研究所 碩 士 論 文., 2010
[7]David Sanchez Mendoza(University Pompeu Fabra), “Emulation Elcectric Guitar Effects With Neural Network” Computer Engineering Graduation Project, 2005
[8]E. Bai. “A blind approach to Hammerstein–Wiener model identification,” Automatica, 38(6):967–979, 2002.
[9]L. Rabiner, “On the use of autocorrelation analysis for pitch detection,” IEEE Transactions on Acoustics, Speech and Signal Processing, vol. 25, no. 1, pp.24-33, 1977