We propose an efficient alignment workflow that leverages a human-in-the-loop approach to perform audio-to-score alignment. This repository contains the code for using this workflow as well as reproducing results from our ISMIR 2024 paper.
Irmak Bukey, Michael Feffer, and Chris Donahue. (in press) "Just Label the Repeats for In-The-Wild Audio-to-Score Alignment." Proceedings of the International Society for Music Information Retrieval Conference (ISMIR), 2024.