“We wanted to put together the technologies that could realize the vision of being able to communicate a message to another person just by thinking about it; basically, some sort of verbal telepathy,” researcher Asier Marzo told Digital Trends. “There are two parts for such a system: Recognizing the message, and transmitting it to another person.”
For recognizing the words, the team attempted to first use sensors to capture the “subvocal electrical signals,” referring to the small electrical signals people send to their muscles and vocal cords when they pronounce words in their mind. Unfortunately, they concluded that this was not feasible. Instead, they based their electromyography (EMG) system around the electrical signals generated by the muscles when people silently mouth words. This proved to give a strong enough readout that it was possible to differentiate between 10 words with more than 80 percent accuracy.
For transmitting the message, they used a phenomenon called “sound through ultrasound,” which involved modulating ultrasound to create an audible sound, but with the high directivity of ultrasound.
“The sender silently mouths a command,” Marzo explained. “Surface electrodes placed over his or her face capture the electrical signals generated by the muscles and a machine learning algorithm identifies the word. Then, a directional speaker emits the command in a very directive manner towards the target. During this process, the beam is finely electronically steered at the target guided by an eye-tracking system or a laser pointer.”
There are various applications for the work, Marzo continued. One use could be for soldiers to have the tech integrated into their helmets. Instead of shouting to communicate, the sender could silently mouth a command such as “take cover” or the location of an enemy and have this message beamed to only the intended recipient. Another may be among divers, which could be especially effective since sound travels much better through water than through air.
Going forward, the goal is to develop the technology so it can detect a broader range of words, as well as shrinking the speaker system to allow it to be surreptitiously built into clothing. (At present, the giant speaker strapped to the wearer’s forehead, and electrodes attached to their mouth, would be a bit of a giveaway that the technology is being used.)
“There are commercial directional speakers available, but they are either too pricey or have bad sound quality and limited teach,” Marzo said. “My intention for the future is to release an Instructables [document] on how to put together a directional speaker using off-the-shelf and cheap components. Also, how to customize it depending on your needs of reach and sound quality.”
