arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

Umulsai:一种用于朝鲜正谱记谱法的纯文本格式和浏览器内排版系统

Umulsai: A Plain-Text Format and In-Browser Engraving System for Jeongganbo, Korean Mensural Notation

Danbinaerin Han

arXiv 2607.24230首次发表:更新:

AI 中文总结

研究针对朝鲜正谱记谱法与西方排版系统不兼容且缺乏实用数字工具的问题,提出了Umulsai编辑器,其核心包含特定纯文本格式、排版引擎和无依赖实现,经专家主导的AI辅助工作流程开发并验证,可实现高质量输出与多种功能。

AI 中文摘要

朝鲜正谱记谱法是15世纪在朝鲜创建的基于网格的有量记谱法,仍用于韩国传统音乐。其结构与西方五线谱排版系统不兼容,且缺乏实用数字工具。本文介绍了Umulsai,一个完全在网页浏览器中运行的静态、无服务器的正谱记谱法编辑器。其核心包括:一种人类可读的纯文本格式,其空间排列与记谱网格一一对应;一个排版引擎,将已出版乐谱的排版惯例形式化为自动规则;一种无依赖实现,使文档保持纯数据,实现打印质量输出、播放以及与最近提出的面向机器的正谱记谱法编码进行互换。我们指定了格式及其节奏语义,描述了排版模型及其排版边缘情况,并讨论了局限性以及与MusicXML/MEI和光学音乐识别管道的计划互操作性。该实现是通过由领域专家领导的人工智能辅助工作流程产生的,并根据已出版的参考乐谱进行了行为验证;本报告中披露了此工作流程及其验证机制。

英文摘要

Jeongganbo is a grid-based mensural notation created in fifteenth-century Korea and still in standard use for Korean traditional music. Its structure -- a vertical column of square beat-cells read right to left, in which the position and share of a syllable inside a cell encode its duration -- is incompatible with engraving systems built for Western staff notation, and no practical digital tool for producing it has been in sustained distribution. This report describes the technical design of Umulsai, a jeongganbo editor that runs entirely in a web browser as a static, serverless application. Its core is (i) a human-writable plain-text format whose spatial arrangement corresponds one-to-one with the notational grid, so that duration is expressed by position and share in the text exactly as on the page; (ii) an engraving engine that formalizes the typesetting conventions of published scores -- right-to-left column flow, daegang section lines, octave-variant Sinographs, ornament attachment and micro-spacing -- as automatic rules; and (iii) a dependency-free implementation in which documents remain plain data, enabling print-quality output, playback, and interchange with recently proposed machine-oriented encodings of jeongganbo. We specify the format and its rhythmic semantics, describe the engraving model and its typographic edge cases, and discuss limitations and planned interoperability with MusicXML/MEI and optical music recognition pipelines. The implementation was produced through an AI-assisted workflow led by a domain expert and verified behaviorally against published reference scores; this workflow and its verification regime are disclosed in the report.

论文原文

arXiv 摘要页 · PDF 原文 · HTML 原文

↑