Rethinking Evaluation Methodology for Audio-to-Score Alignment
This work addresses the need for more precise evaluation in audio-to-score alignment, which is important for researchers and practitioners in music information retrieval, but it is incremental as it focuses on refining existing methodologies rather than proposing new alignment methods.
The paper tackles the problem of evaluating audio-to-score alignment algorithms by providing a formal definition and introducing new evaluation metrics, resulting in a study that compares these metrics with standard ones on classical algorithms using a dataset from KernScores and MAESTRO performances.
This paper offers a precise, formal definition of an audio-to-score alignment. While the concept of an alignment is intuitively grasped, this precision affords us new insight into the evaluation of audio-to-score alignment algorithms. Motivated by these insights, we introduce new evaluation metrics for audio-to-score alignment. Using an alignment evaluation dataset derived from pairs of KernScores and MAESTRO performances, we study the behavior of our new metrics and the standard metrics on several classical alignment algorithms.