Vibration exercise (VE) has been suggested as an effective option to improve muscle strength and power performance. Several studies link the effects of vibration training to enhanced neuromuscular stimulation and typically to involuntary reflex mechanisms. However, the underlying mechanisms are still unclear and information for the most appropriate vibration training protocols is limited. This study proposes to realize a new vibration exercise system for the biceps brachii. Amplitude, frequency, and baseline of the vibrating load, which is generated by an electromechanical actuator, can be adjusted dynamically by a feedback control loop. A second-order model is employed to identify the relation between the mechanical load and the input voltage driving the actuator. An adaptive normalized least mean square algorithm is proposed to remove the motion artifacts from the measured electromyography (EMG) data. Our results show a high correlation (0.99) between the second-order model fit and the measured data, permitting accurate control on the supplied load for vibrations up to 80 Hz. Furthermore, preliminary validation with 4 volunteers showed an excellent performance in the motion artifact removal, enabling reliable evaluation of the neuromuscular activation.