The Role of Individual Difference in Judging Expressiveness of Computer-Assisted Music Performances by Experts

Abstract

Computational systems for generating expressive musical performances have been studied for several decades now. These models are generally evaluated by comparing their predictions with actual performances, both from a performance parameter and a subjective point of view, often focusing on very specific aspects of the model. However, little is known about how listeners evaluate the generated performances and what factors influence their judgement and appreciation. In this article, we present two studies, conducted during two dedicated workshops, to start understanding how the audience judges entire performances employing different approaches to generating musical expression. In the preliminary study, 40 participants completed a questionnaire in response to five different computer-generated and computer-assisted performances, rating preference and describing the expressiveness of the performances. In the second, “GATM” (Gruppo di Analisi e Teoria Musicale) study, 23 participants also completed the Music Cognitive Style questionnaire. Results indicated that music systemizers tend to describe musical expression in terms of the formal aspects of the music, and music empathizers tend to report expressiveness in terms of emotions and characters. However, high systemizers did not differ from high empathizers in their mean preference score across the five pieces. We also concluded that listeners tend not to focus on the basic technical aspects of playing when judging computer-assisted and computer-generated performances. Implications for the significance of individual differences in judging musical expression are discussed

An Interdisciplinary Review of Music Performance Analysis

“…This might indicate that performances are rated based on some form of perceived distance from an 'ideal' per-formance. Apart from music education, assessment of performances is also an important area of focus for the evaluation of computer generated music performances (Bresin and Friberg, 2013) where researchers have primarily focused on listening studies to understand the effect of musical knowledge and biases on rating performances (De Poli et al, 2014) and the degree to which computer generated performances stack up against those by humans (Schubert et al, 2017).…”

Computational Models of Expressive Music Performance: A Comprehensive and Critical Review

“…De Poli et al (2014) and Schubert et al (2014a) specifically study how the audience judges entire performances of computational models, by analyzing listeners' scores of several aspects including technical accuracy, emotional content and coherence of the performed style. The listeners were categorized into two different cognitive styles: music systemizers (those who judge a performance in technical and formal terms) and music empathizers (describe a performance in terms of its emotive content).…”

CaRo 2.0: An Interactive System for Expressive Music Rendering

“…The CaRo 2.0 system won that final stage (see Table 3). The results as well as the comments by the experts are analysed in [41]. The performances played during Rencon 2011 can be listened to in http://smc.dei.unipd.it/advances hci/.…”