Perceptually Motivated Scoring of Musical Meter Classification Algorithms
In this paper, perceived confusions between the four most popular meters 2/4, 3/4, 4/4 and 6/8 in Western music are examined. A theoretical framework for modeling these confusions is proposed and translated into a perceptually motivated objective score that can be used for the evaluation of meter classification algorithms with respect to meter labels that were elicited from a single annotator. Experiments with three artificial and two real algorithms showed that the new score is preferable over the traditional accuracy since the score rewards algorithms that make reasonable errors and seems to be more robust against different annotators.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.