Output details

Learning expressive percussion performance under different visual feedback conditions

This article investigates the effectiveness of the provision of visual feedback for the practicing and improvement of music performance. The task is to perform a repeated musical pattern in a particular style (laid-back, rushed or on the beat). The effectiveness of two types of visual feedback (low-level and high-level) are compared with practicing without visual feedback, using just aural feedback. Although the results are not fully conclusive, there is evidence of the particular benefit of the high level feedback. This type of feedback has the benefit of sensitivity to different ways of performing music. In other words, the system learns from examples what is distinctive to a collection of associated performances and uses this information to provide feedback. The article has important implications for performance research that generally uses simple low-level features to investigate and represent performance.

NOTE: Sconul finds the same article with a wrong ordering of the authors. This is the pre-publication in 2010 and has received 2 citations (5 according to Google scholar).

This research is being continued to further define analysis procedures that can be applied in a diversity of contexts (first design presented at AHRC Workshop: Cross-Disciplinary Perspectives on Expressive Performance, co-authored by Simon Tucker).

The research has been presented at international conferences, including an invited presentation at the University of Leeds (Design, Performance & Technology: Human/Technology Interface).

Alex Brandmeyer was PG student on this project. Timmers was his main supervisor and postdoctoral researcher on the project. Timmers' role concerned in particular data analysis and conceptualisation of the high level feedback. Alex programmed the software, ran most of the experiments and did a considerable part of the writing. Peter Desain and Makiko Sadakata were involved in research design and gave feedback throughout the process.