Tinnitus is a phantom sound that one hears in the ears or in the head without any external source. Sound therapy is one of the most effective techniques for tinnitus treatment that have been proposed. In order to investigate mechanisms of tinnitus generation and the clinical effects of sound therapy, we have proposed computational models with plasticity using a neural oscillator or a neuronal network model described by simplified Hodgkin-Huxley equations. In the present paper, another neuronal network model with a different structure based on a recent physiological consideration is proposed and the simulation results of the model are described. The model is able to replicate the clinical results that human auditory system temporarily halts perception of tinnitus following sound therapy.