Real-time recognition of piano performance for controlling variable acoustic systems

Abstract

With the increasing popularity of multi-use performance halls, variable acoustic systems are in high demand. User interfaces with presets are often used to switch between various acoustical settings in which users select a preset suitable for a particular ensemble type/size before the start of a concert. Challenges are faced, however, as musical note rate often rapidly changes while reverberation time remains static in the performance space. When the note rate of the performance exceeds a certain tempo threshold, the decay of each note begins to blend with the onset of the following notes, causing a masking effect between notes. This masking effect has been shown to cause performers to adapt to the performance space but deviate from the score. In order to reduce the limits presented by static reverberation time and provide the musician with increased dynamic freedom, an adaptive system is presented which more thoroughly implements variable acoustic systems already at play. The system responds in real-time to these changing parameters during a performance, allowing this masking effect to be minimized. When tested on pianists, while no consistent performance trends were discovered, it was found that the adaptive setting was favored over the static wet and dry settings by over half of the subjects.