Facial paralysis is a highly burdening condition, resulting in a patient's inability to move his mimic musculature on one or both sides of his face. This condition compromises the patient's communication and facial expressions, and thus dramatically reduces his quality of life. The current treatment for chronic facial paralysis relies on a complex reconstructive surgery. This publication proposes a novel, less invasive approach for dynamic facial reanimation. The use of a smart material, namely a Dielectric Elastomer Actuator (DEA) is proposed for facial motion restoration, thus avoiding the traditional two-stage free muscle transfer procedure and allowing for a faster recovery of the patient. DEAs are a type of electroactive polymers, showing promising properties similar to natural muscles such as the fact that they are soft, lightweight and allow for large displacements. As a result, a study of the facial muscles and neural interfaces, notably the ones responsible for mouth movement, was performed, in order to implement a realistic setup. In this paper, a non-invasive neural interface based on myoelectric signal is used in order to establish a real-time control of the actuator. Visible motion of a skin model is produced in real time, by synchronizing the actuator to the activity of a healthy muscle, with a maximal delay of 108 ms resulting from the signal processing and a delay of less than 30 ms related to the actuation of the DEA. This shows that the usage of DEA combined with a neural interface presents a promising approach for treatment of facial paralysis.