TY - JOUR
T1 - From Genre Classification to Rhythm Similarity
T2 - Computational and Musicological Insights
AU - Esparza, Tlacael Miguel
AU - Bello, Juan Pablo
AU - Humphrey, Eric J.
N1 - Funding Information:
This work was supported by the National Science Foundation under grant IIS-0844654.
Publisher Copyright:
© 2014, © 2014 Taylor & Francis.
PY - 2015/1/2
Y1 - 2015/1/2
N2 - Traditionally, the development and validation of computational measures of rhythmic similarity in music relies on proxy classification tasks, often equating rhythm similarity to genre. In this paper, we perform a comprehensive, cross-disciplinary exploration of the classification performance of a state-of-the-art system for rhythm similarity. By synthesizing the methods of quantitative analysis with a musicological perspective, detailed insight is gained into the various facets that affect system behaviour, consisting of three main areas: rhythmic sensitivities of a given feature representation, idiosyncrasies of the data used for evaluation, and the tenuous relationship between rhythmic similarity and genre. Through this study, we provide perspective on gauging the abilities of a computational system beyond classification accuracy, as well as a deeper understanding of system design and evaluation methodology as a musically meaningful exercise.
AB - Traditionally, the development and validation of computational measures of rhythmic similarity in music relies on proxy classification tasks, often equating rhythm similarity to genre. In this paper, we perform a comprehensive, cross-disciplinary exploration of the classification performance of a state-of-the-art system for rhythm similarity. By synthesizing the methods of quantitative analysis with a musicological perspective, detailed insight is gained into the various facets that affect system behaviour, consisting of three main areas: rhythmic sensitivities of a given feature representation, idiosyncrasies of the data used for evaluation, and the tenuous relationship between rhythmic similarity and genre. Through this study, we provide perspective on gauging the abilities of a computational system beyond classification accuracy, as well as a deeper understanding of system design and evaluation methodology as a musically meaningful exercise.
KW - audio analysis
KW - information retrieval
KW - machine learning
KW - music analysis
UR - http://www.scopus.com/inward/record.url?scp=84940217667&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84940217667&partnerID=8YFLogxK
U2 - 10.1080/09298215.2014.929706
DO - 10.1080/09298215.2014.929706
M3 - Article
AN - SCOPUS:84940217667
SN - 0929-8215
VL - 44
SP - 39
EP - 57
JO - Journal of New Music Research
JF - Journal of New Music Research
IS - 1
ER -