Harmony Graph, a Social-Network-Like Structure, and Its Applications to Music Corpus Visualization, Distinguishing and Music Generation

Computational Linguistics and Chinese Language Processing Vol. 15, No. 1, March 2010, pp. 1-18 1 © The Association for Computational Linguistics and Chinese Language Processing Harmony Graph, a Social-Network-Like Structure, and Its Applications to Music Corpus Visualization, Distinguishing and Music Generation Wei-An Chen∗, Jihg-Hong Lin+, and Shyh-Kang Jeng# Abstract In this research project, we propose a model, the Harmony Graph, to decompose music into a social-network-like structure according to its harmonies. The whole Harmony Graph network represents the harmony progressions in music. The Harmony Graph is utilized to visualize, distinguish, and generate music for four prepared corpora using social network techniques. We experimented on different characteristics in social network analysis, and we found significant differences among the Harmony Graphs of the four corpora. A new measure called Agglomeration is created to characterize the agglomerating phenomenon that cannot be described sufficiently by existing measures. A corpus-based music composition method is also proposed in this research. By performing random-walk in a Harmony Graph, we generated new music that differs from yet reflects the style of music pieces in the corpus. With the link prediction technique, we also generated music more pleasant aurally than simply using random walks. Keywords: Social Network Analysis, Corpus Visualization, Corpus-Based Generation ∗ Computer Music Laboratory (JCMG), Graduate Institute of Communication Engineering, National Taiwan University E-mail: r98942051@ntu.edu.tw + Business & Marketing Strategy Labortory, Chunghwa Telecommunication Laboratories E-mail: johnlin@cht.com.tw # Department of Electrical Engineering and Graduate Institute of Communication Engineering, National Taiwan University E-mail: skjeng@cc.ee.ntu.edu.tw 2 Wei-An Chen et al. 1. Introduction Corpus is the basis of Corpus Linguistics. It is also important in Computer Music (Manaris et al., 2007). In the research of corpus-based generation, such as generating articles from text corpora (Stribling et al., 2009; Marom & Zukerman, 2005), speech synthesis based on audio corpora (Iida et al., 2003), and generating music from music corpora (Conklin, 2003; Polashek et al., 2005), one interesting topic is that the selection of corpora may lead to results with different styles. For example, an article generated from a corpus of Abraham Lincoln may reveal his style, which reads differently from one generated from a corpus of William Shakespeare. In this paper, we develop a new model, the Harmony Graph, to make music using corpus-based generation and use this model for music-corpus distinction. The Harmony Graph is applied to organize a music corpus into a social-network-like structure, analogous to the Word Graph (Oerder & Ney, 1993) in Corpus Linguistics. Four distinct music corpora were prepared that are collections of music in different genres. In Corpus Linguistics, there has been relevant research on text corpus visualization (Paley, 2002; Fortuna et al., 2005; Rohrer et al., 1998), which generally provides the overall concept of the corpora. Nevertheless, they cannot be used to tell corpora apart at a glance. In the area of Computer Music, distinction of music pieces into different genres using Harmony Graphs is found to be accurate and believed to be new. In addition to visual inspection, we apply social network analysis to …

Content truncated to 500 words