基於GPT-2和近端策略優化的符號音樂生成;Symbolic Music Generation Using GPT-2 and Proximal Policy Optimization

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/96297

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/96297

Title:	基於GPT-2和近端策略優化的符號音樂生成;Symbolic Music Generation Using GPT-2 and Proximal Policy Optimization
Authors:	張翔威;Chang, Hsiang-Wei
Contributors:	資訊工程學系
Keywords:	音樂生成;近端策略優化;強化學習;符號音樂生成;GPT-2
Date:	2025-01-09
Issue Date:	2025-04-09 17:38:00 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究旨在探索和實現基於GPT-2和近端策略優化（PPO）強化學習的符號音樂生成方法。音樂生成在人工智慧與機器學習領域中一直是重要的研究課題之一，目的是通過算法自動生成具有藝術價值和情感表達的音樂作品。本研究通過探討GPT-2在生成旋律、伴奏及和弦方面的應用潛力，並引入PPO強化學習算法，提升生成音樂的質量和多樣性。研究結果表明，結合GPT-2和PPO的符號音樂生成方法能夠有效提升音樂的創造性和多樣性，為音樂創作和應用帶來新的可能性。本研究不僅在理論上具有重要意義，還在實際應用中具有廣泛的前景。;This study aims to explore and implement a symbolic music generation method based on GPT-2 and Proximal Policy Optimization (PPO) reinforcement learning. Music gen eration has long been an important research topic in the fields of artificial intelligence and machine learning, aiming to automatically generate music works with artistic value and emotional expression through algorithms. This research explores the application potential of GPT-2 in generating melodies, accompaniments, and chords, and introduces the PPO reinforcement learning algorithm to improve the quality and diversity of gen erated music. The results show that the symbolic music generation method combining GPT-2 and PPO can effectively enhance the creativity and diversity of music, bringing new possibilities to music creation and application. This research is of great theoretical significance and has broad prospects for practical application.
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	17	View/Open

社群 sharing

Loading...