Characterizing Subtle Timbre Effects of Drum Strokes Played with Different Technique

In instrumental playing, musicians control sound characteristics such as loudness and timbre. While the many years of training for professional musicians ensure that most changes in timbre are intentional, others arise from changes in playing position, technique applied, or timing (Danielsen et. al 2015). Although subtle, such changes in timbre may still be heard and add to the quality of the performance, also for brief, percussive sounds. Drum strokes where the drumstick is allowed to freely rebound from the drum head – “normal” strokes – appear to have different audible quality compared to “controlled” strokes, where a player restrains the stick from freely moving up after the hit (Dahl & Altenmüller, 2008). Although audible, these differences in timbre are not always captured by features traditionally used such as log attack time and temporal centroid. An additional problem is the brevity of percussive sounds, making the use of those descriptors that require frame-based processing (e.g. spectral flux, spectral contrast) difficult (Bigoni & Dahl, 2018). In this context, it is crucial to identify the signal phases (e.g. attack vs. decay or transient vs. steady-state) in a way that is perceptually relevant and meaningful for drum sounds. The goal of this study is to evaluate whether audio descriptors that capture characteristics of the transient part of the waveform differ between Normal and Controlled strokes. Similar to Danielsen et al. (2015), we define this transient part to occur between the onset and the temporal centroid of each stroke, but we also separate the initial attack.