Melodic Skeleton Generation using Simulated Annealing

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/95476

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95476

Title:	Melodic Skeleton Generation using Simulated Annealing
Authors:	雷聿文;Lei, Yu-Wen
Contributors:	資訊工程學系
Keywords:	模擬退火演算法;旋律骨架;風格轉換;相似度比較;生成伴奏;simulated annealing algorithm;melody skeleton;style transformation;similarity comparison;accompaniment generation
Date:	2024-07-13
Issue Date:	2024-10-09 16:53:32 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究提出了一種基於模擬退火演算法的旋律骨架生成方法，旨在開發出一種能夠隨機生成輸入音樂上連貫且有意義的旋律結構之方法，同時保持計算效率。旋律骨架是音樂創作過程中的關鍵基礎，提供了主要的旋律框架和節奏結構，為更複雜的旋律細節發展奠定了基礎。本研究方法利用模擬退火算法在探索廣闊搜索空間及避免局部最佳解的優勢，通過理解旋律中相對重要的音樂元素，並設計評估生成旋律品質的標準來實現旋律骨架的生成。我們詳細介紹了模擬退火算法在本問題上的具體實現，包括解空間的表示、目標函數的設計和退火過程。為驗證所提方法的有效性，我們進行了與原始作曲的相似性比較實驗。結果顯示，基於模擬退火演算法的方法能夠隨機生成符合音樂先驗知識且具有多樣性的高品質旋律骨架，可參考其附錄 1。此外，我們還探索了該方法在自動伴奏生成、風格轉換、數據集創建、音樂理論教育工具和互動音樂生成系統等各個領域的潛在應用。總結來說，本研究引入了一種生成音樂旋律骨架的穩健且創新的方法，利用模擬退火演算法的隨機性和音樂理論的先驗知識，未來具有在自動伴奏生成、風格轉換、數據集創建、音樂理論教育工具和互動音樂生成系統等方面的廣泛應用前景。 ;This study proposes a melody skeleton generation method based on the simulated annealing algorithm, aiming to develop a method that can randomly generate musically coherent and meaningful melodic structures while maintaining computational efficiency. Melody skeletons are a key foundation in the music composition process, providing the primary melodic framework and rhythmic structure, thus laying the groundwork for the development of more complex melodic details. This study′s method leverages the advantages of the simulated annealing algorithm in exploring a vast search space and avoiding local optima, by understanding relatively important musical elements in melodies and designing criteria to evaluate the quality of generated melodies. We provide a detailed explanation of the specific implementation of the simulated annealing algorithm in this problem, including the representation of the solution space, the design of the objective function, and the annealing process. To validate the effectiveness of the proposed method, we conducted similarity comparison experiments with original compositions, the results of which can be referenced in Appendix 1. The results show that the simulated annealing algorithm-based method can randomly generate high-quality melodic skeletons that are consistent with musical prior knowledge and exhibit diversity. Additionally, we explored the potential applications of this method in various fields, such as automatic accompaniment generation, style transformation, dataset creation, music theory education tools, and interactive music generation systems. In conclusion, this study introduces a robust and innovative method for generating musical melody skeletons, utilizing the randomness of the simulated annealing algorithm and iii the prior knowledge of music theory. In the future, this method has broad application prospects in automatic accompaniment generation, style transformation, dataset creation, music theory education tools, and interactive music generation systems.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	21	View/Open

社群 sharing

Loading...