Music Alignment Uncovered: Representations, Algorithms and Hands-On Tools

Abstract

Music lives in the interplay of different representations: from printed sheet music to machine-readable symbolic representations like MEI or MusicXML to audio and video recordings. For each of these representations, there could exist multiple versions that correspond to the same musical material (e.g., different performances of the same piece, different editions of the same work). Automatic music alignment is a task in the context of Music Computing and Music Information Research (MIR) that focuses on automatically linking elements from one representation to another. Automatic music alignment is a fundamental component of different applications, from comparative analysis of music performances to interactive applications such as automatic accompaniment systems and visualizations.

In this half-day workshop we will provide a practical, hands-on overview of music alignment. The workshop will consist of a comprehensive overview of alignment technologies for both symbolic music and audio, including a practical presentation of tools, datasets and formats for representing alignments. Rather than focusing primarily on the technical aspects of alignment algorithms, we will focus on practical aspects of using alignment for the purpose of music analysis in different contexts, showcasing pipelines, open-source tools and applications.

Repository

All materials are available on GitHub.