Parameter Optimisation for Woodwind Single-Reed Models


Autoria(s): Chatziioannou, Vasileios; Van Walstijn, Maarten
Data(s)

01/09/2010

Resumo

Time-domain modelling of single-reed woodwind instruments usually involves a lumped model of the excitation mechanism. The parameters of this lumped model have to be estimated for use in numerical simulations. Several attempts have been made to estimate these parameters, including observations of the mechanics of isolated reeds, measurements under artificial or real playing conditions and estimations based on numerical simulations. In this study an optimisation routine is presented, that can estimate reed-model parameters, given the pressure and flow signals in the mouthpiece. The method is validated, tested on a series of numerically synthesised data. In order to incorporate the actions of the player in the parameter estimation process, the optimisation routine has to be applied to signals obtained under real playing conditions. The estimated parameters can then be used to resynthesise the pressure and flow signals in the mouthpiece. In the case of measured data, as opposed to numerically synthesised data, special care needs to be taken while modelling the bore of the instrument. In fact, a careful study of various experimental datasets revealed that for resynthesis to work, the bore termination impedance should be known very precisely from theory. An example is given, where the above requirement is satisfied, and the resynthesised signals closely match the original signals generated by the player.

Formato

application/pdf

Identificador

http://pure.qub.ac.uk/portal/en/publications/parameter-optimisation-for-woodwind-singlereed-models(bbf8e75d-7937-4de3-aa53-ba307badb240).html

http://pure.qub.ac.uk/ws/files/18205943/Parameter.pdf

Idioma(s)

eng

Direitos

info:eu-repo/semantics/openAccess

Fonte

Chatziioannou , V & Van Walstijn , M 2010 , Parameter Optimisation for Woodwind Single-Reed Models . in Proceedings of the Second Vienna Talk . pp. 43-46 .

Tipo

contributionToPeriodical