P4-14: Counterpoint Error-Detection Tools for Optical Music Recognition of Renaissance Polyphonic Music
Thomae Elias, Martha E*, Cumming, Julie, Fujinaga, Ichiro
Subjects (starting with primary): Applications -> music heritage and sustainability ; MIR fundamentals and methodology -> symbolic music processing ; MIR tasks -> optical music recognition ; Domain knowledge -> computational music theory and musicology ; Applications -> digital libraries and archives
Presented Virtually: 4-minute short-format presentation
This paper discusses part of a larger project to preserve and increase access to Guatemalan music sources written in mensural notation by using a digitization and music information retrieval (MIR) workflow to obtain both digital images and symbolic scores with editorial corrections. The workflow involves MIR tools such as optical music recognition (OMR), automatic voice alignment for mensural notation, editorial correction software, and computational counterpoint error detection.
In this paper, we evaluate whether the use of automatic counterpoint error-detection tools makes the correction process more efficient. The results confirm that marking illegal dissonances in the score following the rules of Renaissance counterpoint indeed makes the process of editorial correction of scribal errors in Renaissance music more efficient by reducing the time taken and improving the accuracy of such corrections. Moreover, marking the illegal dissonances in the score also allowed us to catch OMR errors that had passed through undetected at a previous stage of the workflow.