CNPM Database:一个用于计算音乐学的中国民族五声调式数据库  

CNPM Database:A Chinese National Pentatonic Mode Database for Computational Musicology

作　　者：任伟鑫 车明锦 汪照文 孟文武 李沁雨 胡佳弋 夏凡[2] 李伟[4,5] REN Weixin;CHE Mingjin;WANG Zhaowen;MENG Wenwu;LI Qinyu;HU Jiayi;XIA Fan;LI Wei(Department of Composition,Sichuan Normal University,Chengdu,Sichuan 610066,China;College of Experimental Art,Sichuan Conservatory of Music,Chengdu,Sichuan 610021,China;Department of Music AI and Information Technology,Central Conservatory of Music,Beijing 100031,China;School of Computer Science and Technology,Fudan University,Shanghai 200438,China;Shanghai Key Laboratory of Intelligent Information Processing,Fudan University,Shanghai 200433,China)

机构地区：[1]四川师范大学作曲系,四川成都610066 [2]四川音乐学院实验艺术学院,四川成都610021 [3]中央音乐学院音乐人工智能与音乐信息科技系,北京100031 [4]复旦大学计算机科学技术学院,上海200438 [5]复旦大学上海市智能信息处理重点实验室,上海200433

出　　处：《复旦学报（自然科学版）》2022年第5期555-563,共9页Journal of Fudan University：Natural Science

基　　金：国家重点研发计划(2019YFC1711800);国家自然科学基金(62171138)。

摘　　要：中国民族五声调式数据库(CNPM Database)的建设参照世界音乐数据库建设中手工标注与计算机相结合的工作思路,收集、存储并标注“宫、商、角、徵、羽”5种调式(含五声、六声、七声)的音频,同时找到应用场景与技术模型,为中国民族音乐特征的分析与检索提供原始数据。该数据库是第一个可用于计算音乐学的五音数据库,提出针对音频的中国民族调式自动识别系统的设计思路,分别使用基于模板匹配和基于卷积神经网络的方法进行了实验,得到分类结果。自动识别方法对于调式主音音名(12种)、调式样式(5种)和调式所属同宫系统(12种)的分类较为准确,但对调式种类(6种)的识别率还需提高。使用计算机代替人工筛选的过程可以极大扩充中国民族调式音乐数据库,以便其应用于多种场景。The construction of Chinese National Pentatonic Mode database(CNPM Database)is based on the idea of combining manual labeling and computer in the construction of the world music database,by collecting,storing and labeling the audio in five modes of Gong,Shang,Jue,Zhi and Yu,including pentatonic,hexatonic and heptatonic.At the same time,it finds the application scenarios as well as technical models,and provides original data for the analysis and retrieval of the features of Chinese national music.The database is the first five tone database that can be used for computational musicology.The design idea of the automatic Chinese national mode recognition system for audio is proposed.Experiments are carried out using template matching method and CNNbased method respectively,and the classification results are obtained.The automatic recognition method is more accurate in classifying the mode tonic(12 categories),the mode style(5 categories)and the TongGong system to which the mode belongs(12 categories),but the recognition rate of mode types(6 categories)needs to be improved.Using computer to replace the manual screening process can greatly expand the national mode music database so that it can be used in a variety of scenarios.

关 键 词：中国民族五声调式 数据库 音乐信息检索 五音疗法 

分 类 号：J61[艺术—音乐]