Piano Touch Analysis: a Matlab Toolbox for Extracting Performance Descriptors from High Resolution Keyboard and Pedalling Data

Abstract

We hereby present the analytic tools developed as a MATLAB toolbox for exploring the most subtle features of piano touch and gesture. From high-sample-rate, high- precision precision key, hammer and pedal tracking data about a performance gathered thanks to the Bösendorfer grand piano-embedded CEUS digital recording system, the toolbox main functions can extract exhaustive features detailing the pianist’s touch as a thorough account of nu- ances in articulation, timing, dynamics, attack and ped- alling. Each performance and gestural control thereof is thus described over each of its notes and chords. By com- paring several performances, it is possible to characterize the gestural control of expression in piano performance, as the correlations among piano touch features towards one examined expressive parameter. The analytic functions in the toolbox — with piano touch feature visualization and comparison, chords and notes selection tools, score- performance matching and advanced, automated statisti- cal analyses and visualization thereof — allow for rig- orous, quantitative exploration of expressive performance and its gestural control, which here is especially applied towards investigating the use of timbre as expressive de- vice in piano performance. With these tools, we thus in- tend to build a gestural mapping of piano timbre. 1.