PIANO PERFORMANCE EVALUATION DATASET WITH MULTI-LEVEL PERCEPTUAL FEATURES

Jisoo Park, Jongho Kim, Jeong Mi Park, Ahyeon Choi, Wen-Syan Li, Jonghwa Park, seung-won hwang

Keywords: dataset, music, perception, piano performance evaluation, hierarchical attention network

TL;DR: We introduce a piano performance evaluation dataset with multi-level perceptual features and expert annotations, along with a baseline model.

Abstract:

This study aims to build a comprehensive dataset that enables the automatic evaluation of piano performances. In real-world piano performance, especially within the realm of classical piano music, we encounter a vast spectrum of performance variations. The challenge lies in how to effectively evaluate these performances. We must consider three critical aspects: 1) It is essential to gauge how performers
perceive and express, and listeners perceive the music, not just the compositional characteristics of music such as beat stability measured from a metronome. 2) Beyond fundamental elements like pitch and duration, we must also embrace higher-level features such as interpretation. 3) Such evaluation should be done by experts to discern the nuanced performances. Regrettably, there exists no dataset
that addresses these challenging evaluation tasks. Therefore, we introduce a pioneering dataset PercePiano, annotated by music experts, with more extensive features capable of representing these nuanced aspects effectively. It encapsulates piano performance with a wide range of perceptual features that are recognized by musicians. Our evaluation benchmark includes a novel metric designed to
accommodate the inherent subjectivity of perception. For the baseline models, we pinpoint a significant issue in current transformer-based models. We in response introduce a new baseline that leverages hierarchical levels of performance, which shows results comparable to that of large pre-trained models. In conclusion, our research opens new possibilities for comprehensive piano performance evaluation.

Primary Area: datasets and benchmarks

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 5327