Developed by researchers at the Massachusetts Institute of Technology (MIT), the AlterEgo headpiece uses electrodes that scan the wearer’s jaw and face to figure out what they are thinking about saying. When we think about vocalizing words, but don’t actually speak them, the muscles in our faces produce neuromuscular signals in a process known as subvocalization. The headset has been trained to translate these signals into about 20 basic words, and its developers believe that it could learn a lot more with additional training.

The AlterEgo device also uses bone conduction technology, sending auditory signals to the inner ear via vibrations in the skull and jawbones, to speak to the wearer. The entire communication process is, therefore, silent.

Check out the device in action in the video below.