This paper addresses rub defect loudspeaker detection. The study includes a simulation with a rub model based on classical static coulomb friction added to the loudspeaker nonlinearities parametric model to demonstrate the electric current signal viability to rub failure detection. The electric current signal is analyzed by means of Zhao-Atlas-Marks distribution (ZAMD). A failure extractor based on relevant ZAMD frequency regions segmentation and Mahalanobis distance is presented. The simulation and experimental results show the goodness and reliability of rub detection method presented.