以音樂為主題的遊戲一直是相當受歡迎的遊戲類型，從大型遊戲機台、電視遊戲主機、電腦遊戲，到近年來蓬勃發展的手持裝置，都可以看見音樂遊戲的蹤影。音樂打擊遊戲，其所使用的譜面須經由人工標記產生。本論文提出一套程序化的方法，從偵測聲音起始時間得到的資料當中，篩選出可用的、相對重要的時間點，並對它們加以分類，產生型別與難度，最後整合這些結果並產生譜面。透過此方法製作的音樂遊戲，譜面經由自動化過程產生，大幅提升產能；另一方面，使用者能即時將所持有音樂轉換成譜面，增加遊戲豐富性。

Music game is a very popular game genre all the time. From arcade game machines, TV game consoles and PC games to recently booming mobile handsets, the presence of music games is in sight. Onset-based music games, which use notes to play, need the notes to be tagged by humans. This thesis proposes a system of procedural method, which selects useful and relatively important time points from the data yielded by onset detection, categorizes them into types and difficulty levels, and finally integrates these results into playable notes. The proposed system generates notes automatically, which massively enhances the productivity. On the other hand, users may transform their own music into notes in real-time, introducing much interest to the game.

摘要 i
Abstract ii
List of Figures v
List of Tables vii
1 Introduction 1
2 Related Work 4
2.1 Onset Detection 4
2.1.1 Temporal Energy 4
2.1.2 Spectral Energy 4
2.1.3 Spectral Flux 5
2.1.4 Spectral Difference 6
2.1.5 Phase Deviation 6
2.1.6 Complex Domain 7
2.1.7 Kullback-Leibler 8
2.2 Harmonic Percussive Sound Separation 9
2.3 Procedural Content Generation 11
3 System 15
3.1 Problem Definition 15
3.2 Harmonic Percussive Sound Separation 17
3.3 Onset Detection 17
3.3.1 Weighted Spectral Flux 17
3.3.2 Peak-Picking 19
3.4 Result Combination 21
3.5 Nearby Onsets Removal 22
3.6 Grouping 24
3.6.1 Type Determination 24
3.6.2 Difficulty Determination 26
3.7 Content Generation 28
4 Performance Evaluation 29
4.1 Methodology 29
4.2 Dataset 30
4.3 Results and Discussion 30
4.3.1 Experiment: Comparison of Onset Detection Functions 30
4.3.2 Experiment: Search for Optimal Frequency Weighting 32
4.3.3 Experiment: Search for Optimal Peak-Picking Threshold 34
4.3.4 Experiment: Benefits of Sound Separation 35
4.3.5 Experiment: Comparison of Difficulty Levels 36
4.3.6 Experiment: Comparison of Approaches to Type Determination 41
5 Conclusions and Future Work 44
5.1 Conclusions 44
5.2 Future Work 44
References 46
Appendices 48
Result of Nelder-Mead Simplex Method 48

[1]J. P. Bello, L. Daudet, S. Abdallah, C. Duxbury, M. Davies and M. B. Sandler, “A tutorial on onset detection in music signals,” IEEE Transactions on Speech and Audio Processing, vol. 13, no. 5, pp. 1035-1047, September 2005.
[2]S. Dixon, “Onset detection revisited,” in Proceedings of the International Conference on Digital Audio Effects, 2006.
[3]P. Masri, “Computer modeling of sound for transformation and synthesis of musical signal,” Ph.D. dissertation, University of Bristol, Bristol, U.K., 1996.
[4]S. Hainsworth and M. Macleod, “Onset detection in musical audio signals,” in Proceedings of the International Computer Music Conference, 2003.
[5]P. M. Brossier, “Automatic annotation of musical audio for interactive applications,” Ph.D. dissertation, Queen Mary University of London, London, U.K., 2006.
[6]N. Ono, K. Miyamoto, J. Le Roux, H. Kameoka and S. Sagayama, “Separation of a monaural audio signal into harmonic/percussive components by complementary diffusion on spectrogram,” in Proceedings of the European Signal Processing Conference, 2008.
[7]J. Togelius, G. N. Yannakakis, K. O. Stanley and C. Browne, “Search-based procedural content generation: A taxonomy and survey,” IEEE Transactions on Computational Intelligence and AI in Games, vol. 3, no. 3, pp. 172-186, September 2011.
[8]“Echo,” n.d., in Wikipedia, retrieved June 29, 2014, from
http://en.wikipedia.org/wiki/Echo
[9]“Takahashi Meijin,” n.d., in Wikipedia, retrieved June 29, 2014, from
http://en.wikipedia.org/wiki/Takahashi_Meijin
[10]J. C. Lagarias, J. A. Reeds, M. H. Wright and P. E. Wright, “Convergence properties of the Nelder-Mead simplex method in low dimensions,” SIAM Journal on Optimization, vol. 9, no. 1, pp. 112-147, December 1998.
[11]“Equal-loudness contour,” n.d., in Wikipedia, retrieved June 29, 2014, from
http://en.wikipedia.org/wiki/Equal-loudness_contour