Indian researcher develops device that understands what you are thinking
Indian-origin researcher Arnav Kapur, along with his team at the MIT Media Lab, has developed a device that can understand and act upon the words you are saying in your head. The computer interface in the gadget, called AlterEgo, can transcribe the words that you subvocalize but don't speak out loud, enabling you to silently converse with computing devices without any movement at all.
What is subvocalization?
Subvocalization is the practice of saying words in your head, most commonly observed during silent reading. While it is deemed as a hindrance in the act of speed reading, subvocalization might actually change the way we interact with AI assistants and mobile devices.
How AlterEgo understands what you are saying in your head
The prototype white gadget wraps around the back of the neck and touches the face at seven locations along the jaw with curved appendages. The wearable's electrodes pick up and measure the neuromuscular signals in the jaw that are triggered by internal verbalization. The signals are then fed to a machine-learning system that relates them to corresponding words. This is how it understands you.
How AlterEgo responds to your queries
Now for being responsive, the device has a pair of bone-conduction headphones that convey information through the bones of the face to the inner ear, without obstructing a user's ear canal/auditory experience. Thus, the device forms an entirely silent ecosystem where you can give it commands through subvocalization and it can respond without engaging any of your physical senses.
You can silently ask the time, add up grocery bill
AlterEgo has been tested for functions like basic addition, asking the time, and navigating a video streaming platform through subvocalization. The system has been used in a chess game where it was silently fed with the opponents' moves and in turn, it gave computer-recommended responses.
Researchers want AlterEgo to understand full-on conversations
The device can perform limited computational tasks as of now because it has been trained with about 20 words. However, researchers are improving AlterEgo on the go, training it for more elaborate conversations. "We're in the middle of collecting data, and the results look nice. I think we'll achieve full conversation some day," Kapur said.