运用计算方法研究京剧音乐 Computational Methods for Jingju Music Research

在这个讲座中,我们介绍了在CompMusic项目中的京剧研究。在这里可以找到相关的资料。

In this presentation, we introduce the research on jingju carried out in the CompMusic project. Here you can find the related materials.

在北京大学的讲座视频，由北京大学中国音乐学社主办（2015年5月23日）
Video of the presentation at Peking University, hosted by the Peking University Chinese Music Institute (May 23, 2015) (Youtube) (优酷)

幻灯片 Slides (中文) (English)

演示视频 Demo videos:

• 自动分段 Automatic segmentation (下载 download)
• 主旋律提取 Predominant melody extraction (下载 download)
• 颤音识别 Vibrato detection (下载 download)
• 唱词与音频对齐 Lyrics to audio alignment (下载 download)
• 自动MIDI记谱 Automatic transcription in MIDI (下载 download)
• Dunya (演示 demo)

讲座概要

音乐信息检索（Music Information Retrieval，简称MIR）是一个融合了计算机科学，数字信号处理，音乐学，人工智能等学科的新兴研究方向。在MIR兴起的过去20年里，随着这个领域的快速发展，智能化音乐技术得到广泛应用。然而目前大多数的MIR研究主要以西方流行音乐为对象，而对非西方传统音乐的处理和分析有待提高。为了填补这个空白，推动音乐信息技术在世界各音乐文化中的应用，由欧洲研究理事会出资的CompMusic 项目（http://compmusic.upf.edu/zh-hans）于2011年启动。这个项目以西班牙庞培法布拉大学音乐技术研究组为中心，其目的在于针对世界五种文化的传统音乐，研发音乐分析和处理的计算工具，包括以京剧音乐为样本的中国音乐。

此次讲座将由 CompMusic 项目的总负责人Xavier Serra教授，以及该项目两名研究员，Rafael Caro和张朔介绍他们的研究成果并讨论音乐信息检索技术在京剧音乐分析和处理中的应用。CompMusic 的目标之一在于探索不同文化的传统音乐给MIR和音乐学分析带来的新的可能性。讲座中将涉及到其中以京剧为对象的诸多研究课题，包括音乐结构／段落的自动划分，定弦和主音识别，京剧行当的唱腔风格分析与比较（颤音，音色，强度），唱词和音频以及乐谱和音频的自动对齐，节奏与速度分析，唱腔的自动板式划分，以及锣鼓经的自动识别等等。同时，这些技术在宏观音乐分析探索方面也有广泛的应用前景，包括在京剧音乐数据库中提取各类统计信息，以及开发以音乐内容特征为基础的音乐浏览器工具。

Xavier Serra是西班牙庞培法布拉大学信息与通信技术系的副教授。他同时是该系音乐技术研究组的总监。Serra教授于1989年在美国斯坦福大学取得计算机音乐技术博士学位。他的博士论文开创了以确定性和随机频谱合成／分析／转化音乐声音信号的研究，成为业内有重要影响的文献。他的研究方向包括音乐信号的分析，描述，以及合成，兼并基础研究和实际应用，并融合科学技术和人文艺术领域的研究课题。近年来，Serra教授活跃于西班牙和国际声音和音乐计算领域的前沿。作为MIR领域诸多核心期刊的评审，他多次受邀于各国学术机构和国际会议，就MIR的现状和未来的挑战进行讲学。Serra教授于2011年获得欧洲研究理事会高级研究基金，以开展旨在拓展多元文化音乐计算的为期五年的CompMusic研究项目。

Rafael Caro Repetto 是西班牙庞培法布拉大学音乐技术研究组CompMusic项目博士生研究员。他在CompMusic项目中主要负责以MIR的方法分析京剧的声腔 （导师：Xavier Serra, Manel Ollé)。他的学术背景主要包括音乐学和东亚研究。Rafael Caro 拥有伦敦大学亚非学院的民族音乐学硕士学位。他的主要研究方向是以京剧为重点的中国传统音乐。

张朔是西班牙庞培法布拉大学音乐技术研究组CompMusic项目的合作研究员。他同时是美国乔治城大学语言学系计算与理论语言学博士候选人。他的研究方向主要包括计算语言学－语音学认知模型，自然语言处理，语料库,和音乐信息检索。目前张朔参与的研究项目与软件开发包括CompMusic 项目中京剧音乐和语音声调信号的时间序列数据发掘, 以及德国柏林洪堡大学和美国乔治城大学合作开发的可视化语料库搜索Java网络应用软件，ANNIS.张朔毕业于北京大学环境科学与工程学院以及美国匹兹堡大学音乐系。

Abstract of the presentation

The young discipline of Music Information Retrieval (MIR) has experienced a remarkable growth and development during its two decades of existence. However, most of its research has focused on contemporary western commercial music, what might cause its results to be culturally biased. The aim of the project CompMusic: Computational Models for the discovery of the world's music (http://compmusic.upf.edu) is to tackle this bias and at the same time to benefit from the new research problems posed by different music traditions. The CompMusic project, funded by the European Research Council, and carried out at the Music Technology Group (Universitat Pompeu Fabra, Barcelona), applies a data driven, culturally specific approach to the research in five music traditions, including Chinese jingju music.

In this presentation, CompMusic's general PI, Prof. Xavier Serra, and two of its researchers, Rafael Caro and Zhang Shuo, will explain the potential of computational methods for jingju music research. Some of the tasks that can benefit from these methods include automatic segmentation of jingju arias into musically relevant sections, computation of tuning and first degree, analysis and comparison of different role-types' styles of singing, including features such as vibrato, timbre, and loudness, lyrics and score to audio alignment, rhythm and tempo analysis, segmentation of arias into different banshi, and percussion patterns recognition. A further benefit of this approach is the possibility of applying these analyses to large collections of recordings. This allows the extraction of statistical information about the collection, or developing browsing tools based on musical features.

Xavier Serra is Associate Professor of the Department of Information and Communication Technologies and Director of the Music Technology Group at the Universitat Pompeu Fabra in Barcelona. After a multidisciplinary academic education he obtained a PhD in Computer Music from Stanford University in 1989 with a dissertation on the spectral processing of musical sounds that is considered a key reference in the field. His research interests cover the analysis, description and synthesis of sound and music signals, with a balance between basic and applied research and approaches from both scientific/technological and humanistic/artistic disciplines. Dr. Serra is very active in promoting initiatives in the field of Sound and Music Computing at the local and international levels, being involved in the editorial board of a number of journals and conferences and giving lectures on current and future challenges of the field. He has recently been awarded an Advanced Grant of the European Research Council to carry out the project CompMusic aimed at promoting multicultural approaches in music computing research.

Rafael Caro Repetto is a researcher in the CompMusic project, where he carries out his PhD research on jingju shengqiang, under the supervision of Prof. Xavier Serra and Prof. Manel Ollé. His formal training is in Musicology and East Asian Studies. He holds a MMus in Ethnomusicology from the School of Oriental and African Studies (University of London). His main research area is Chinese traditional music, with a special focus on jingju.

Zhang Shuo is a collaborator/researcher for the CompMusic project. He is also a PhD candidate in Computational and Theoretical Linguistics at the Department of Linguistics, Georgetown University (Washington DC). His research spans computational linguistics and music information retrieval. He currently works on time-series data mining of the music-speech melodic contours in Beijing opera, and the ANNIS Java Web Application (open-source tool for the search and visualization of speech corpora). Shuo obtained his BS degree from College of Environmental Sciences and Engineering, Peking University and his MA degree from Department of Music, University of Pittsburgh.