Shaping the behaviour of feedback instruments with complexity-controlled gain dynamics

Feedback instruments offer radical new ways of engaging with instrument design and musicianship. They are defined by recurrent circulation of signals through the instrument, which give the instrument 'a life of its own' and a 'stimulating uncontrollability'. Arguably, the most interesting musical behaviour in these instruments happens when their dynamic complexity is maximised, without falling into saturating feedback. It is often challenging to keep the instrument in this zone; this research looks at algorithmic ways to manage the behaviour of feedback loops in order to make feedback instruments more playable and musical; to expand and maintain the `sweet spot'. We propose a solution that manages gain dynamics based on measurement of complexity, using a realtime implementation of the Effort to Compress algorithm. The system was evaluated with four musicians, each of whom have different variations of string-based feedback instruments, following an autobiographical design approach. Qualitative feedback was gathered, showing that the system was successful in modifying the behaviour of these instruments to allow easier access to edge transition zones, sometimes at the expense of losing some of the more compelling dynamics of the instruments. The basic efficacy of the system is evidenced by descriptive audio analysis. This paper is accompanied by a dataset of sounds collected during the study, and the open source software that was written to support the research.