Abstract:
Typical physical modeling synthesis methods for brass instruments contain a lip dynamics model and an acoustic flow model for the lips. Both elements contain nonlinearities, leading to a nonlinear feedback loop. Problems that arise with this type of system include difficulty in predicting the pitch of the synthesized tone and lack of robust methods for getting tones to ``speak.'' Elimination of the lip dynamics model leads to replacement of the nonlinear feedback system with a simpler nonlinear filtering system. In this model the lip reed is replaced by a zero impedance regulating valve which moves at the frequency of the desired output, according to measured lip data. Whereas for the conventional synthesis model, the frequency of synthesized sound is neither the resonant frequency of the lips nor the resonant frequency of the bore; in the case of the driven lip model, the synthesized frequency is exactly the same as the frequency at which the lip valve is driven. The driven model also guarantees that synthesized tones will speak, because amplitude of lip oscillation becomes a direct control parameter. Since harmonic generation in brass instruments is not directly caused by lip dynamics, spectrally rich tones can be generated by this model.