Due to various circumstances, a considerable number of people are no longer able to express themselves in a normal, audible voice. Traditional methods of restoring communication for these patients present numerous problems, mainly because they merely create a poor-quality replacement voice rather than restoring the individual's natural voice.

The MyVoice project, led by PhD. Eng. Michael Wand, Senior Researcher at IDSIA, in collaboration with Prof. Tanja Schultz of the University of Bremen, develops a new and innovative speech generation device based on facial muscle activity to restore the patient's natural voice. It is expected that such a device will also be useful when discreet or confidential communication is needed in public places. The project is funded by the Swiss National Science Foundation (SNSF) and the German Research Foundation (DFG).

The method is based on recording the user's facial muscle activity using light electrodes (surface electromyography). The signal is then interpreted by a state-of-the-art artificial intelligence system and converted into audible voice. In this way, even if the user moves his or her lips completely silently, the original voice can be recovered. The main goal of the project is to make the user's experience of the system as pleasant as possible, particularly by minimizing the time needed to adapt the system to a new user.

«With this project we can provide real help to patients with speech difficulties, who are often still excluded from daily interaction with other people. We also have the exciting scientific challenge of developing an artificial intelligence system that interacts with the user in real time and evolves over time following specific needs. We believe such systems will become increasingly important in the future,» Wand explains.