Using Deep Learning for Text to Asia Music Generation
Artificial Intelligence, especially Deep Learning had become the trend for generating text, image, video, and music. Music generation or automatic composition by AI was available for a while. Most music tools generate western music, which is well developed for a long history. However, Asian music theory could have different focuses, although the music was influenced by western music. When small chamber music is composed for Asia music, it takes a few steps to construct melody and accompany. We combine Deep Learning techniques with traditional algorithms, to develop a web-based system for public usage. Five types of Asia music were developed. The result is appreciated by professional musicians. In this article, we discuss the system components and deep learning modules. We believe that, music rules can be incorporated in reinforcement learning. And, different Asia music accompany, including percussion instruments, can be created following the melody generated by reinforcement learning. The completed system is available at http://140.115.51.218:8888. The tool was used by musicians (not the authors) for generating music and performed officially by professionals.
Authors
- Pe-Chung Chen
- Yu-Cheng Lin
- Timothy K. Shih (ORCID: https://orcid.org/0000-0003-4154-4752)
- Cheng-Han Wu
Institutions
- National Central University (TW)
- Yuan Ze University (TW)
Publication Details
- Published
- 2024-09-30
- DOI
- https://doi.org/10.1109/aixheart62327.2024.00008
- Primary Topic
- Music and Audio Processing
- Type
- article
- Field-Weighted Citation Impact
- 0.00