Estimation of Clarinet Reed Parameters by Inverse Modelling

Analysis of the acoustical functioning of musical instruments invariably involves the estimation of model parameters. The broad aim of this paper is to develop methods for estimation of clarinet reed parameters that are representative of actual playing conditions. This presents various challenges because of the difficulties of measuring the directly relevant variables without interfering with the control of the instrument. An inverse modelling approach is therefore proposed, in which the equations governing the sound generation mechanism of the clarinet are employed in an optimisation procedure to determine the reed parameters from the mouthpiece pressure and volume flow signals. The underlying physical model captures most of the reed dynamics and is simple enough to be used in an inversion process. The optimisation procedure is first tested by applying it to numerically synthesised signals, and then applied to mouthpiece signals acquired during notes blown by a human player. The proposed inverse modelling approach raises the possibility of revealing information about the way in which the embouchure-related reed parameters are controlled by the player, and also facilitates physics-based re-synthesis of clarinet sounds.