請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/85612完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 雷欽隆(Chin-Laung Lei) | |
| dc.contributor.author | Ting-Hao Chang | en |
| dc.contributor.author | 張庭豪 | zh_TW |
| dc.date.accessioned | 2023-03-19T23:19:42Z | - |
| dc.date.copyright | 2022-07-08 | |
| dc.date.issued | 2022 | |
| dc.date.submitted | 2022-06-29 | |
| dc.identifier.citation | [1] OSU!, https://osu.ppy.sh/home, accessed June 2021 [2] Schluter Jan, Bock Sebastian. “Musical onset detection with convolutional neural networks”, 6th International workshop on machine learning and music, Prague, Czech Republic, 2013. [3] Juan Pablo Bello, Laurent Daudet, Samer Abdallah, Chris Duxbury, Mike Davies, Mark B.Sandler. “A tutorial on onset detection in music signals”, IEEE transactions on speech and audio processing, vol.13, no.5, pp.1035-1047, 2005. [4] Schluter Jan, Bock Sebastian. “Improved musical onset detection with convolutional neural networks”, IEEE international conference on acoustics, speech and signal processing, pp. 6979-6983, 2014. [5] Rongfeng Li, Yijun Chen. “Score generation for taiko no tatsujin based on machine learning”, IEEE conference on multimedia information processing and retrieval, pp. 408-413, 2020. [6] Nogaj A.F. “A genetic algorithm for determining optimal step patterns in dance dance revolution”, Technical report, State University of New York at Fredonia, 2005. [7] Emily Halina, Matthew Guzdial. “Taikonation: Patterning-focused chart generation for rhythm action games”, arXiv:2107.12506v1, 2021. [8] Honglak Lee, Peter Pham, Yan Largman, and Andrew Ng. “Unsupervised feature learning for audio classification using convolutional deep belief networks”, Advances in neural information processing systems 22, pp.1096-1104, 2009. [9] Roland Memisevic, Christopher Zach, Marc Pollefeys, Geoffrey E Hinton. “Gated softmax classification”, Advances in neural information processing systems 23, pp. 1603-1611, 2010. [10] Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov. “Dropout: a simple way to prevent neural networks from overfitting”, The journal of machine learning research 15, pp. 1929–1958, 2014. [11] Chris Donahue, Zachary C. Lipton, Julian McAuley. “Dance Dance Convolution”, arXiv:1703.06891v3, 2017. [12] Eck, Douglas. “A first look at music composition using lstm recurrent neural networks”, Technical Report IDSIA-07-02, 2002. [13] Greff, Klaus, Srivastava, Rupesh K, Koutnik, Jan, Steune-brink, Bas R, Schmidhuber, Jurgen. “Lstm: A search space odyssey”, IEEE transactions on neural networks and learning systems, pp. 2222-2232, 2016. [14] Zaremba, Wojciech, Sutskever, Ilya, Vinyals, Oriol. “Recurrent neural network regularization”, arXiv:1409.2329, 2014. [15] Yubin Liang, Wanxiang Li, Kokolo Ikeda. “Procdeural content generation of rhythm games using deep learning methods”, Entertainment computing and serious games, pp. 134-145, 2019. [16] Erik Marchi, Giacomo Ferroni, Florian Eyben, Leonardo Gabrielli, Stefano Squartini, Bjorn Schuller. “Multi-resolution linear prediction based features for audio onset detection with bidirectional LSTM neural networks”, IEEE international conference on acoustics, speech and signal processing, pp. 2164-2168, 2014. [17] Florian Eyben, Sebastian Bock, Bjorn Schuller, Alex Graves. “Universal onset detection with bidirectional long short-term memory neural networks”, International society for music information retrieval conference, 2010. [18] Buda Mateusz, Maki Atsuto, Mazurowski Maciej A.. “A systematic study of the class imbalance problem in convolutional neural networks”, arXiv:1710.05381v2, 2018. [19] Jont B.Allen, Lawrence R.Rabiner. “A unified approach to short-time Fourier analysis and synthesis”, Proceedings of the IEEE, pp.1558-1564, 1977. [20] Philippe Hamel, Yoshua Bengio, Douglas Eck. “Building musically-relevant audio features through multiple timescale representations”, International society for music information retrieval conference, 2012. [21] Diederik P.Kingma, Jimmy Ba. “Adam: A method for stochastic optimization”, arXiv preprint arXiv:1412.6980, 2014. [22] https://ai.ntu.edu.tw/resource/handouts/ML9-1.html [23] https://www.itread01.com/content/1543994346.html | |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/85612 | - |
| dc.description.abstract | 太鼓達人是世界上流行的節奏遊戲之一，它模擬了音樂下的太鼓演奏模式。然而，在這個遊戲之中存在一個煩人的系統，就是節奏點的判斷與安排。而這件事往往需要耗費大量的人力跟時間去完成，才得以讓遊戲譜面跟歌曲形成一致的遊玩模式，而本篇論文則是研發可以自動產生遊戲譜面的方式。 本篇論文設計了兩個步驟，第一是時間點的生成步驟，它決定了何時應該放置節拍。第二是節奏點類型的生成步驟，它決定了第一步驟的時間點上所生成的節奏類型。 在這項研究中，有使用一些特殊的方法。首先是模糊標籤(fuzzy label)的使用，因為訓練數據極度不平衡，模糊標籤可以使其變得更平滑，也可以提高數據的含標籤量，促使數據變得足夠平衡。再來是讓數據的移動幅度與快速傅立葉轉換的窗口大小相符合，如此一來，就不會出現數據在前處理的過程中被分成兩份。最後是使用數據量更大的卷積長短期記憶(C-LSTM)模型。 在測試數據上面，我們將預測時間點步驟的F-score，在六十四分之一拍的前提下，從0.7769提高到0.8312，在三十二分之一拍的前提下，從0.8765提高到0.9068。 | zh_TW |
| dc.description.abstract | Taiko no Tatsujin is one of the popular rhythm games in the world. It simulates playing the taiko drum in time with music. However, this game exists a system struggle, that is to replicate human-like patterning. The placement of game objects in relation to each other to form congruent patterns based on events in the song. This thesis introduces two steps to generate the beatmaps. First step is timestamp generation, it decides when to put the beat. Second step is action type generation, it decides what to put on the time in first step. In this research, some special methods are purposed. First is “fuzzy label”, since the training data is extreme unbalance. Fuzzy label can make the changing smoother and increase the positive rate of data, which make the data balance enough to train. Second is make the stride fit the window lengths of the music. Which the training data will not be cut into two part by unfit stride. Third is the C-LSTM model with larger data size. On the test data, the thesis improved the F-score of timestamp prediction from 0.7769 to 0.8312 for 64 semi-notes and 0.8765 to 0.9068 for 32 semi-notes. | en |
| dc.description.provenance | Made available in DSpace on 2023-03-19T23:19:42Z (GMT). No. of bitstreams: 1 U0001-2406202209100400.pdf: 2397671 bytes, checksum: 98a573e5c4166cf4a1c5a9c29aaff746 (MD5) Previous issue date: 2022 | en |
| dc.description.tableofcontents | CONTENTS 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 1.1 Research Scope 2 1.2 Contribution 2 Chapter 2 Background 4 2.1 Flow Chart 4 2.2 Songs audio source and limit 5 2.3 Data used for training 5 2.3.1 Timestamp Generation 5 2.3.2 Action Type Generation 6 Chapter 3 Related Work 7 3.1 Improved Musical Onset Detection With Convolutional Neural Networks 7 3.2 Score Generation for Taiko no Tatsujin based on Machine Learning 8 3.3 TaikoNation: Patterning-focused Chart Generation for Rhythm Action Games 9 3.4 Dance Dance Convolution 10 3.5 Procedural Content Generation of Rhythm Games Using Deep Learning Methods 10 Chapter 4 Preparation of Training Data 13 4.1 Number of collected data 13 4.2 Preprocess the data to spectrogram 13 4.3 Transform Spectrogram into Training Data 14 4.4 Original Data to Fuzzy Label Data 14 Chapter 5 Timestamp generation 16 5.1 Multiple research model and dataset 16 5.2 Training parameter and method 19 5.3 Method for analyzing the results and reason 19 Chapter 6 Action Type Generation 28 6.1 Model and dataset 28 6.2 Training parameter and method 29 6.3 Method for analyzing the results and reason 30 Chapter 7 Conclusions 31 REFERENCE 32 LIST OF FIGURES Fig 1.1 Interface of the Taiko no Tatsujin 1 Fig 2.1 Flow Chart 4 Fig 2.2 OSU website 5 Fig 2.3 Timestamp generation data 6 Fig 2.4 Action type generation data 6 Fig 3.1 Visualize spectrogram process 7 Fig 3.2 CNN architectures used in this thesis 8 Fig 3.3 C-LSTM architectures used in this thesis 9 Fig 3.4 Visualize two steps for beatmap generation in this thesis 11 Fig 3.5 Fuzzy label in this thesis 12 Fig 4.1 Short-Time Fourier Transform 13 Fig 4.2 Data with and without overlap 14 Fig 4.3 Visualize data transform through fuzzy label 15 Fig 5.1 CNN based model graph 16 Fig 5.2 C-LSTM model graph 17 Fig 5.3 C-LSTM large model graph 18 Fig 5.4 The operation process of Adam 19 Fig 5.5 The operation process of Binary Cross Entropy 19 Fig 5.6 The way threshold work 20 Fig 5.7 AUC example 21 Fig 6.1 Example of input features and output 28 Fig 6.2 LSTM model graph 29 Fig 6.3 The operation process of Categorical Cross Entropy 30 LIST OF TABLES Table 3.1 Measure method in this thesis 10 Table 5.1 Detail of all the dataset in this research 18 Table 5.2 True positive False positive table 21 Table 5.3 Detail of accuracy and AUC 24 Table 5.4 Result for validation part under 64 semi-notes 24 Table 5.5 Result for testing part under 64 semi-notes 25 Table 5.6 Result for testing part with validation threshold under 64 semi-notes 25 Table 5.7 Result for validation part under 32 semi-notes 26 Table 5.8 Result for testing part under 32 semi-notes 26 Table 5.9 Result for testing part with validation threshold under 32 semi-notes 27 Table 6.1 Taiko Don Ka True positive False positive table 29 Table 6.2 Result for action type generation 30 | |
| dc.language.iso | en | |
| dc.subject | 深度學習 | zh_TW |
| dc.subject | 音樂 | zh_TW |
| dc.subject | 節奏 | zh_TW |
| dc.subject | 太鼓達人 | zh_TW |
| dc.subject | 深度學習 | zh_TW |
| dc.subject | 機器學習 | zh_TW |
| dc.subject | 音樂 | zh_TW |
| dc.subject | 節奏 | zh_TW |
| dc.subject | 太鼓達人 | zh_TW |
| dc.subject | 機器學習 | zh_TW |
| dc.subject | Taiko no Tatsujin | en |
| dc.subject | Music | en |
| dc.subject | Rhythm | en |
| dc.subject | Machine Learning | en |
| dc.subject | Music | en |
| dc.subject | Deep Learning | en |
| dc.subject | Deep Learning | en |
| dc.subject | Taiko no Tatsujin | en |
| dc.subject | Rhythm | en |
| dc.subject | Machine Learning | en |
| dc.title | 透過深度學習自動生成太鼓達人遊戲譜面 | zh_TW |
| dc.title | Automatic Generation of Taiko Games Using Deep Learning Methods | en |
| dc.type | Thesis | |
| dc.date.schoolyear | 110-2 | |
| dc.description.degree | 碩士 | |
| dc.contributor.oralexamcommittee | 郭斯彥(Sy-Yen Kuo),顏嗣鈞(Hsu-Chun Yen) | |
| dc.subject.keyword | 機器學習,深度學習,太鼓達人,節奏,音樂, | zh_TW |
| dc.subject.keyword | Machine Learning,Deep Learning,Taiko no Tatsujin,Rhythm,Music, | en |
| dc.relation.page | 34 | |
| dc.identifier.doi | 10.6342/NTU202201087 | |
| dc.rights.note | 同意授權(全球公開) | |
| dc.date.accepted | 2022-07-01 | |
| dc.contributor.author-college | 電機資訊學院 | zh_TW |
| dc.contributor.author-dept | 電機工程學研究所 | zh_TW |
| dc.date.embargo-lift | 2022-07-08 | - |
| 顯示於系所單位： | 電機工程學系 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| U0001-2406202209100400.pdf | 2.34 MB | Adobe PDF | 檢視/開啟 |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。