Developing and evaluating a hybrid wind instrument

A hybrid wind instrument generates self-sustained sounds via a real-time interaction between a computed excitation model (such as the physical model of human lips interacting with a mouthpiece) and a real acoustic resonator. Attempts to produce a hybrid instrument have so far fallen short, in terms of both the accuracy and the variation in the sound produced. The principal reason for the failings of previous hybrid instruments is the actuator which, controlled by the excitation model, introduces a fluctuating component into the air flow injected into the resonator. In the present paper, the possibility of using a loudspeaker to supply the calculated excitation signal is evaluated. A theoretical study has facilitated the modeling of the loudspeaker-resonator system and the design of a feedback and feedforward filter to successfully compensate for the presence of the loudspeaker. The resulting self-sustained sounds are evaluated by a mapping of their sound descriptors to the input parameters of the physical model of the embouchure, both for sustained and attack sounds. Results are compared with simulations. The largely coherent functioning confirms the usefulness of the device in both musical and research contexts.