JosquIntab: A Dataset for Content-based Computational Analysis of Music in Lute Tablature

An enormous corpus of music for the lute, spanning some two and half centuries, survives today. Unlike other musical corpora from the same period, this corpus has undergone only limited musicological study. The main reason for this is that it is written down exclusively in lute tablature, a prescriptive form of notation that is difficult to understand for non-specialists as it reveals little structural information. In this paper we present JosquIntab, a dataset of automatically created enriched diplomatic transcriptions in MIDI and MEI format of 64 sixteenth-century lute intabulations, instrumental arrangements of vocal compositions. Such a dataset enables large-scale content-based computational analysis of music in lute tablature hitherto impossible. We describe the dataset, the mapping algorithm used to create it, as well as a method to quantitatively evaluate the degree of arrangement (goodness of fit) of an intabulation. Furthermore, we present two use cases, demonstrating the usefulness of the dataset for both music information retrieval and musicological research. We make the dataset, the source code, and an implementation of the mapping algorithm, runnable as a command line tool, publicly available.