Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Six Dragons Fly Again: Reviving 15th-Century Korean Court Music with Transformers and Novel Encoding

Danbinaerin Han, Mark Gotham, Dongmin Kim, Hannah Park, Sihun Lee, Dasaem Jeong

TL;DR

This work tackles reviving 15th-century Korean court music encoded in Jeongganbo by building a machine-readable dataset via optical music recognition and training transformer-based models for ensemble generation. It introduces a Jeonggan-like encoding with a beat-counter that aligns with Jeongganbo structure, and a BERT-like masked language model for transforming melodies, enabling a complete orchestration from a monophonic heritage melody. Objective metrics (Length Match Rate, F1-Score) and expert reviews show that the proposed encoding and refinement strategies yield ensemble scores that closely resemble the target Yeominlak style while handling limited data. A web demo and a performed premiere demonstrate practical impact for cultural preservation and computational ethnomusicology, with broader implications for encoding-aware generation of traditional music.

Abstract

We introduce a project that revives a piece of 15th-century Korean court music, Chihwapyeong and Chwipunghyeong, composed upon the poem Songs of the Dragon Flying to Heaven. One of the earliest examples of Jeongganbo, a Korean musical notation system, the remaining version only consists of a rudimentary melody. Our research team, commissioned by the National Gugak (Korean Traditional Music) Center, aimed to transform this old melody into a performable arrangement for a six-part ensemble. Using Jeongganbo data acquired through bespoke optical music recognition, we trained a BERT-like masked language model and an encoder-decoder transformer model. We also propose an encoding scheme that strictly follows the structure of Jeongganbo and denotes note durations as positions. The resulting machine-transformed version of Chihwapyeong and Chwipunghyeong were evaluated by experts and performed by the Court Music Orchestra of National Gugak Center. Our work demonstrates that generative models can successfully be applied to traditional music with limited training data if combined with careful design.

Six Dragons Fly Again: Reviving 15th-Century Korean Court Music with Transformers and Novel Encoding

TL;DR

This work tackles reviving 15th-century Korean court music encoded in Jeongganbo by building a machine-readable dataset via optical music recognition and training transformer-based models for ensemble generation. It introduces a Jeonggan-like encoding with a beat-counter that aligns with Jeongganbo structure, and a BERT-like masked language model for transforming melodies, enabling a complete orchestration from a monophonic heritage melody. Objective metrics (Length Match Rate, F1-Score) and expert reviews show that the proposed encoding and refinement strategies yield ensemble scores that closely resemble the target Yeominlak style while handling limited data. A web demo and a performed premiere demonstrate practical impact for cultural preservation and computational ethnomusicology, with broader implications for encoding-aware generation of traditional music.

Abstract

We introduce a project that revives a piece of 15th-century Korean court music, Chihwapyeong and Chwipunghyeong, composed upon the poem Songs of the Dragon Flying to Heaven. One of the earliest examples of Jeongganbo, a Korean musical notation system, the remaining version only consists of a rudimentary melody. Our research team, commissioned by the National Gugak (Korean Traditional Music) Center, aimed to transform this old melody into a performable arrangement for a six-part ensemble. Using Jeongganbo data acquired through bespoke optical music recognition, we trained a BERT-like masked language model and an encoder-decoder transformer model. We also propose an encoding scheme that strictly follows the structure of Jeongganbo and denotes note durations as positions. The resulting machine-transformed version of Chihwapyeong and Chwipunghyeong were evaluated by experts and performed by the Court Music Orchestra of National Gugak Center. Our work demonstrates that generative models can successfully be applied to traditional music with limited training data if combined with careful design.
Paper Structure (24 sections, 6 figures, 2 tables)

This paper contains 24 sections, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Jeongganbo Dataset
  4. Jeongganbo Notation
  5. Machine Readable Dataset
  6. Jeonggan-like Encoding
  7. Other Possible Encodings
  8. Orchestral Part Generation
  9. Transformer Sequence-to-sequence Model
  10. Beat Counter
  11. Experiment and Results
  12. Training
  13. Evaluation Metrics
  14. Length Match Rate
  15. F1-Score
  16. ...and 9 more sections

Figures (6)

  • Figure 1: Overview of the proposed research framework
  • Figure 2: An example of Jeongganbo in the original notion (below) and a broadly equivalent conversion to Western classical notion (above). Dashed lines are part of neither notation and added simply to clarify the temporal alignment between the two systems.
  • Figure 3: Jeonggan-like encoding position labels
  • Figure 4: Comparison between encoding schemes
  • Figure 5: Orchestral part generation
  • ...and 1 more figures