Semi‑Automatic Pipeline for the Transcription of Mensural Polyphony into Symbolic Interpreted Scores

This paper presents the first semi-automatic pipeline for transcribing early polyphonic music, written in separate parts, into symbolic scores. While current optical music recognition (OMR) software can recognize and encode musical symbols, it falls short of producing complete transcriptions from these historical sources. Polyphonic music from the late Middle Ages and Renaissance is written in mensural notation, where note duration cannot be inferred from graphical symbols alone. Accurate interpretation requires consideration of additional features, such as mensuration signs and the contextual placement of notes. This article introduces an integrated MIR pipeline that transcribes digital images of mensural sources into aligned symbolic scores by combining OMR, a mensural rhythmic interpretation expert system, and error-checking technologies—thereby reducing human intervention. A central aim of this work is to improve interoperability among existing specialized tools rather than develop new monolithic solutions and to rely on free, open-access software to enable adoption by institutions and scholars worldwide, including those in developing countries. The pipeline was successfully applied to the Guatemalan choirbook GCA-Gaha 1, demonstrating its practical utility and effectiveness.