| 研究生: |
陳秉萱 Ping-Hsuan Chen |
|---|---|
| 論文名稱: |
基於 CycleGAN 之古箏音樂風格轉換 Music Conversion for Chinese Guzheng Using CycleGAN |
| 指導教授: |
張寶基
Po-Chi Chang |
| 口試委員: | |
| 學位類別: |
碩士 Master |
| 系所名稱: |
資訊電機學院 - 通訊工程學系在職專班 Executive Master of Communication Engineering |
| 論文出版年: | 2020 |
| 畢業學年度: | 108 |
| 語文別: | 中文 |
| 論文頁數: | 38 |
| 中文關鍵詞: | 古箏 、深度學習 、音樂風格轉換 |
| 相關次數: | 點閱:11 下載:0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
隨著科技日新月異,人們之間的距離因此縮短,文化的交流讓世界成為地球村,促使東西方的文化融合更加快速。在現代人的消遣娛樂當中,自從臺灣華語流行歌曲男歌手周杰倫與歌詞創作家方文山合力創作出《娘子》、《東風破》、《菊花台》等歌曲後,受到熱烈的回響,娛樂產業便刮起了一陣「中國風」。由於中國風的興起,更出現了西方樂曲被翻唱成中國風格的作品,翻唱(Cover)一直是樂壇中盛行之事,是由另外一位作者重新詮釋原作音樂,此種作為不僅限於歌唱,亦適用於跨樂器的詮釋,俗稱翻奏。
此篇論文,主要是提出一個基於圖像人工智慧的技術,來成就更好的音樂風格轉換,並且搭上時下流行的中國風,將音樂轉換成中國古箏的版本。透過人類感知實驗評估此方式的結果,獲得4.3的高分(0分為最低分,5分為最高分)。
Chinese Guzheng music is popular from ancient Qin dynasty. However, from the study of modern Chinese history, Chinese people continued to pursue westernization. The Chinese Guzheng also faces the drastic changes of the external environment such as the needs of modern national orchestras. Due to changes, the Guzheng is reformed in the direction of expanding the sound range, increasing the volume. Although new design for the Chinese Guzheng instrument has been proposed, it is too hard for players to catch the music transition because of long time producing and learning new skills.
To improve this issue, we propose music conversion with CycleGAN which is a deep learning technique that involves the automatic training of image-to-image translation models without paired examples. Moreover, this method provides a simple way which can be used by whom never learn any instrument. This is a brand new research with good results.
[1] 何孟翰,數位音樂創作中編曲配器之探討:以〈塵煙〉一曲為例談古風歌創作,https://hdl.handle.net/11296/dyu3mv,2018。
[2] 趙麗妍,當代台灣國語翻唱歌曲現象研究(1990~2011),2012。
[3] 林育嬋,改作權問題之研究,https://hdl.handle.net/11296/pxahsy,2014。
[4] 張晏萍,新型式古箏製作研究,https://hdl.handle.net/11296/4c8bne,2016。
[5] 黃金燕,古箏演奏法在箏樂流派之運用,2014。
[6] 李怡瑩,說唱音樂在傳統箏曲上的應用-以山東箏派及河南箏派為例,2018。
[7] 黃暉閔,當代陝西箏派作品之研究─以《姜女淚》、《雲裳訴》、《秦土情》為例,
2013。
[8] 許紫庭,浙江箏派的演奏技法對當代箏曲創作的影響,https://hdl.handle.net/11296/2n8f2f,1998。
[9] 洪涔綾,從《柳青娘》各家版本演奏譜探討潮州箏樂的彈奏藝術,https://hdl.handle.net/11296/j46aaq,2000。
[10] 張榮麗,訓練深度學習網路之探討,https://hdl.handle.net/11296/53dtbs,2017。
[11] S.Hochreiter, J.Schmidhuber, “Long short-term memory Neural computation,” 9(8):1735–1780, 1997.
[12] Photosounder. Retrieved from http://photosounder.com/
[13] Audio joiner. Retrieved from https://audio-joiner.com/
[14] J.-Y. Zhu, T. Park, P. Isola, and A. A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks,” Proc. Int. Conf. Computer Vision. [Online], Available: https://arxiv.org/abs/1703.10593, 2017.
[15] G. Brunner, A. Konrad, Y. Wang, and R. Wattenhofer, “MIDIVAE: Modeling Dynamics and Instrumentation of Music with Applications to Style Transfer,” arXiv :1809.07600, 2018.
[16] Si. Huang, Q. Li, C. Anil, X. Bao, S. Oore, and R. B. Grosse, “TimbreTron: A WaveNet(CycleGAN(CQT(Audio))) Pipeline for Musical Timbr Transfer,”arXiv:1811.09620, 2018.