The electroacoustic resonator is an efficient electro-active device for noise attenuation in enclosed cavities or acoustic waveguides. It is made of a loudspeaker (the actuator) and one or more microphones (the sensors). So far, the desired acoustic behaviour, expressed in terms of a linear-time-invariant relationship between sound pressure and vibrational motion (the acoustical impedance), has been more efficiently achieved by a model-inversion strategy which is implemented by driving the electrical current in the loudspeaker coil, based upon the measured pressure. The corrector transfer function is defined in the Laplace domain and digitally executed by the classical infinite-impulse-response technique, though a state-space representation could be employed. In this work, we are interested in enforcing a nonlinear behaviour at low sound excitation levels, where the electroacoustic resonator would normally behave as a linear-timeinvariant system. Hence, in order to transform its acoustical response from linear to nonlinear, the model-inversion technique must be reformulated in time domain. The state-space representation of the relationship between the input measured pressure and the output electrical current gives the right perspective and the solution to this problem. We provide the conception of this modelinversion control algorithm capable of transforming a linear-time-invariant acoustical response to potentially any causal acoustical response of the electroacoustic resonator. Such control strategy is tested by targeting a Duffing acoustical response with tunable parameters. Both numerical simulations and experimental tests in quasi-open field validate the approach. The results provided in this contribution open the doors for conceiving non-conventional absorbers which can exploit nonlinear phenomena for noise mitigation even at low excitation amplitudes.