Viscoelastic physical modeling synthesis and machine learning analysis of Jazz drum head tapering

Jazz drummers often damp their snare drums or tom-toms considerably to shorten the drum sound, allowing the performance of complex rhythmic textures and designing the drum timbre. Tapering involves the use of materials such as Mylar or other plastics, attached to the drumheads using gaffa or similar strong adhesive tapes, applied at different positions and in varying amounts. This leads to a considerable change in the eigenmode spectrum of the drumhead and a complex frequency-dependent viscoelastic damping, which is the main reason for the tapering. Using a viscoelastic physical modeling algorithm, a parameter space of possible drum head tapering is computed by varying position, size, and kind of tapered material. Using deep neural networks, this parameter space is trained and used to identify real tapered drum heads concerning the materials. The presented methodology enables the estimation of tapering in historical drum sets from Jazz recordings, leading to a general understanding of the techniques used to design a desired sound. Additionally, it can assist Jazz drummers in selecting particular tapering methods to achieve a desired drum timbre.