RECONSIDERING QUANTIZATION IN MIR Jordan Lenchitz College of Music, Florida State University jlenchitz@fsu.edu ABSTRACT This paper presents a critique of the ubiquity of boilerplate quantizations in MIR research relative to the paucity of en- gagement with their methodological implications. The wide-ranging consequences of reflexivity on the future of scholarly inquiry combined with the near-universal con- temporary recognition of the need to broaden the scope of MIR research invite and merit critical attention. To that end, focusing primarily on twelve-tone equal-tempered pitch and dyadic rhythm models, we explore the practical, cultural, perceptual, historical, and epistemological conse- quences of these pervasive quantizations. We analyze sev- eral case studies of meaningful and successful past re- search that balanced practicality with methodological va- lidity in order to posit several best practices for both future intercultural studies and research centered on more nar- rowly constructed corpora. We conclude with a discussion of the dangers of solutionism on the one hand and the self- fulfilling prophecies of status quoism on the other as well as an emphasis on the need for intellectual honesty in met- atheoretical discourse. 1. INTRODUCTION Fostering cultural diversity in MIR research is not merely a question of “adapting” existing methodologies devel- oped on the basis of certain a priori assumptions to reper- toires or tasks that perhaps challenge said assumptions. In particular, it should also entail critical reflection on the benefits and drawbacks of those assumptions in studies of all repertoires, even (if not especially) those on the basis of which such assumptions were made in the first place. Twelve-tone equal-tempered quantizations of pitch and dyadic quantizations of rhythm represent arguably the most ubiquitous such assumptions in contemporary MIR research, and yet despite their prevalence their founda- tional role underlying many diverse methodological ap- proaches all too often passes unannounced. This paper therefore presents an analysis not of any musical infor- mation in particular but rather of the apparatuses we use to retrieve it from musics in the hopes of fostering productive future methodological conversations. Quantization may be defined as the organization of di- mensional information into discrete sets of values (other- wise known as categorization in perception science), and it is cognitively essential for creating music. That being said, the centrality of the process of quantization to music does not justify the perpetuation of reliance on any “de- fault” quantizations. Indeed, certain quantizations are ubiquitous, all-too-often unstated a priori assumptions un- derlying a substantial majority of MIR methodologies. Are they compromises? Almost invariably yes they are. Do they represent pragmatic choices? Quite possibly they do, depending on the context. But pragmatic compromises or otherwise, the foundational position of these quantizations endows them with consequences that merit consideration in the context of any methodological decision-making and especially if meaningful progress is to be made in expand- ing the purview of MIR and its applications. This paper is structured as follows: after this introduc- tion providing the rationale for a reconsideration of quan- tization in MIR, its consequences practical, cultural, per- ceptual, historical, and epistemological are each explored in turn with an emphasis on pitch and rhythm in the second section. The third section provides illustrative case studies that demonstrate strategies for balancing pragmatism with methodological validity and posits best practices for both intercultural and intracultural research. The fourth and fi- nal section argues for the necessity of avoiding both solu- tionism on the one hand and status quoism on the other and emphasizes the importance of intellectual honesty in met- atheoretical discourse. 2. CONSEQUENCES OF QUANTIZATION 2.1 Practical Consequences The most obvious practical consequence of the ubiquity of certain quantizations in MIR is the widespread availability of platforms built around them and the corresponding no- ticeable absence of alternatives. In and of itself this is un- surprising and not necessarily a drawback; and yet, it can bring about the existence of unfortunate self-fulfilling prophecies. Consider the case of the justifiably popular Py- thon audio analysis package librosa [1], in which one may specify the number of notes per octave but the assumption that they are equally spaced is not so easily changed. As maintainer Brian McFee put it with respect to pitch on the Music Information Retrievers Slack in July 2020, equal temperament “is a compromise, but one I’m willing to live with for the time being; extending to support just intona- tion in a fully consistent way would be a huge undertaking, much bigger than just adding notation support.” We are in- clined to agree with McFee in his assessment that imple- menting meaningful support for tuning systems other than equal temperaments would be a very nontrivial task. At the same time, however, such tasks tend to be welcomed by MIR researchers as motivation for innovation. Why, then, does this remain unaddressed? © J. Lenchitz. Licensed under a Creative Commons Attribu- tion 4.0 International License (CC BY 4.0). Attribution: J. Lenchitz, “Reconsidering Quantization in MIR”, in Proc. of the 22 nd Int. Society for Music Information Retrieval Conf., Online, 2021. 374