EMelodyGen Dataset (ABC notation 旋律情感数据集)

EMOPIA 数据集是一个集合了丰富情感表达的MIDI格式音乐作品的集合，旨在支持音乐情感分析和音乐信息检索领域的研究。通过对EMOPIA数据集中的MIDI文件进行深入的数据处理，我们成功地将所有曲目的第一声部（V1声部）转换成了ABC符号表示法。ABC符号是一种用纯文本表示音乐记谱法的系统，它以简洁明了的方式编码音乐信息，便于计算机处理和分析。在转换过程中，我们特别注意保留了每首曲目的情感标签，这些标签对于理解音乐的情感内容至关重要。由此得到的新数据集不仅为音乐情感分析提供了一个结构化和易于访问的资源，同时也为音乐教育、音乐理论的研究以及自动音乐生成等领域的应用提供了可能。emo_rough4q 是一个综合性的音乐研究资源，它集合了来自多个知名音乐数据库——包括irishman-xml、Wikifonia、ESAC、诺丁汉、Chorale Bach和ccmusic中的乐谱。该数据集利用music21库的analyze方法进行了自动化标注，以提取和识别乐谱中的ABC大小调模式。经过严格的数据筛选和处理流程，最终形成了一个包含345万条记录的大型数据集。这一资源不仅为音乐理论研究、音乐教育和历史音乐作品的分析提供了丰富的素材，也为音乐信息检索、计算机辅助作曲以及音乐情感分析等领域的研究提供了强有力的数据支持，推动了音乐学和音乐人工智能领域的进一步发展。

The EMOPIA dataset is a collection of MIDI-format musical works rich in emotional expressions, designed to support research in the fields of music emotion analysis and music information retrieval. Through in-depth data processing on the MIDI files in the EMOPIA dataset, we successfully converted the first voice (Voice 1, V1) of all tracks into ABC notation. ABC notation is a plain-text musical notation system that encodes musical information in a concise and clear manner, facilitating computer processing and analysis. During the conversion process, special attention was paid to preserving the emotional labels of each piece, which are critical for understanding the emotional content of the music. The resulting new dataset not only provides a structured and easily accessible resource for music emotion analysis, but also enables applications in fields such as music education, music theory research, and automatic music generation.

The emo_rough4q is a comprehensive music research resource that aggregates sheet music from multiple well-known music databases, including irishman-xml, Wikifonia, ESAC, Nottingham, Chorale Bach, and ccmusic. This dataset uses the analyze method of the music21 library for automated annotation to extract and identify major and minor tonal patterns in the sheet music. After strict data filtering and processing workflows, a large-scale dataset containing 3.45 million records was finally formed. This resource not only provides abundant materials for music theory research, music education, and analysis of historical musical works, but also offers strong data support for research in fields such as music information retrieval, computer-aided composition, and music emotion analysis, promoting further development in the fields of musicology and musical artificial intelligence.

搜集汇总

背景与挑战

背景概述

EMelodyGen是一个基于ABC记谱法的旋律情感数据集，包含四个子集：EMOPIA和VGMIDI子集由MIDI文件转换并增强情感标签，Analysis子集提供统计分析以指导数据增强，Rough4Q子集则整合多个数据集并自动标注情感。该数据集旨在支持音乐情感生成研究，包含丰富的统计特征和相关性分析。