Traditional Western musical instruments have evolved to be robust and predictable, responding consistently to the same player actions with the same musical response. Consequently, errors occurring in a performance scenario are typically attributed to the performer and thus a hallmark of musical accomplishment is a flawless musical rendition. Digital musical instruments often increase the potential for a second type of error as a result of technological failure within one or more components of the instrument. Gestural instruments using machine learning can be particularly susceptible to these types of error as recognition accuracy often falls short of 100%, making errors a familiar feature of gestural music performances. In this paper we refer to these technology-related errors as system errors, which can be difficult for players and audiences to disambiguate from performer errors. We conduct a pilot study in which participants repeat a note selection task in the presence of simulated system errors. The results suggest that, for the gestural music system under study, controlled increases in system error correspond to an increase in the occurrence and severity of performer error. Furthermore, we find the system errors reduce a performer’s sense of control and result in the instrument being perceived as less accurate and less responsive.
Log In to access the entire content