This study describes the application of the feedback linearization neural networks, known from neural network computing, to the problem of gradient preemphasis.