Linking Sheet Music and Audio - Challenges and New Approaches

Score and audio files are the two most important ways to represent, convey, record, store, and experience music. While score describes a piece of music on an abstract level using symbols such as notes, keys, and measures, audio files allow for reproducing a specific acoustic realization of the piece. Each of these representations reflects different facets of music yielding insights into aspects ranging from structural elements (e. g., motives, themes, musical form) to specific performance aspects (e. g., artistic shaping, sound). Therefore, the simultaneous access to score and audio representations is of great importance. In this paper, we address the problem of automatically generating musically relevant linking structures between the various data sources that are available for a given piece of music. In particular, we discuss the task of sheet musicaudio synchronization1 with the aim to link regions in images of scanned scores to musically corresponding sections in an audio recording of the same piece. Such linking structures form the basis for novel interfaces that allow users to access and explore multimodal sources of music within a single framework. As our main contributions, we give an overview of the state-of-the-art for this kind of synchronization task, we present some novel approaches, and indicate future research directions. In particular, we address problems that arise in the presence of structural differences and discuss challenges when applying optical music recognition to complex orchestral scores. Finally, potential applications of the synchronization results are presented. 1998 ACM Subject Classification H.5.1 Multimedia Information Systems, H.5.5 Sound and Music Computing, I.5 Pattern Recognition, J.5 Arts and Humanities–Music