The reduction of the wear process at the moment of movement of rail vehicles on curved parts of tracks is an important task. To reduce the amount of wear, different technologies have been applied including an application for the active steering of rail vehicle wheelsets. Active control systems use the same control strategies, such as wheelset/bogie yaw moment control and wheelset/track lateral position control. However, these strategies are not optimal for all types of rail vehicle steering. Therefore, the purpose of this research is to obtain preliminary results for the further development of a strategy that allows the prediction of the angle of attack between the wheel and rail based on noise analysis. A series of experiments were made with a test rig and the obtained data was recorded for different values of the angle of attack and studied by analyzing the generated spectrograms for each sound. Results of this study are presented and the effectiveness of this work is discussed.