Visualizing Air Drums: Analysis of Motion and Vocalization Data Related to Playing Imaginary Drums

Abstract

Air drums, or imaginary drums, are commonly played as a form of participating in musical experiences. The gestures derived from playing air drums can be acquired using accelerometers and then mapped into sound control responses. Commonly, the mapping process relies on a peak-picking procedure that maps local maxima or minima to sound triggers. In this work, we analyzed accelerometer and audio data comprising the motion of subjects playing air drums while vocalizing their expected results. Our qualitative analysis revealed that each subject produced a different relationship between their motion and the vocalization. This suggests that using a fixed peak-picking procedure can be unreliable when designing accelerometer-controlled drum instruments. Moreover, user-specific personalization can be an important feature in this type of virtual instrument. This poses a new challenge for this field, which consists of quickly personalizing virtual drum interactions. We made our dataset available to foster future work in this subject.