A new device from MIT researchers can effectively read users’ thoughts to generate hands-free, voice-free, digital commands, according to a report from Live Science. The AlterEgo headset uses a neural network to translate “subvocalization” patterns, which are small movements in the muscles of your face, larynx, and jaw, that occur when most people read or think of a word. These internal verbalizations are imperceptible to the human eye, but the “augmented intelligence” headset makes use of these signals, integrating the human mind with a digital platform in an unprecedented way. The result is a tool similar to a silent version of Apple’s Siri or Amazon’s Alexa.
According to Arnav Kapur, one of the MIT media lab graduate students involved in the project, and the lead author of a paper on the device:
“Our idea was: Could we have a computing platform that’s more internal, that melds human and machine in some ways and that feels like an internal extension of our own cognition?”
Electrodes on the headset, placed closely against the user’s face, pick up the small movements that occur when the user thinks of a word, or sentence. The device uses Bluetooth to send those signals to a computer, which then uses a neural network to translate the thoughts into commands. This works similarly to a speech-to-text program, but without the need for audible vocalization.
The AlterEgo uses no earbuds or headphones, instead utilizing “bone conduction headphones” that generate their own minute vibrations into the facial bones and ultimately the inner ear. This allows the device to “respond” to your commands and results in what is essentially a silent dialogue between the user and the device. This occurs without obstructing the user’s ability to listen to, or communicate with others.
The prototype was tested on a few tasks with simple commands, limited to about 20 words each, such as arithmetic and chess. The headphones require calibration for each individual user, since everyone’s subvocalization patterns are a bit different.
Early studies into the efficacy of the device have shown an accuracy of 92 percent, in a study of 10 volunteers silently reading a list of randomized numerical digits. Researchers say this accuracy will likely improve with regular use. Users spent 15 minutes calibrating the headset for the arithmetic task.
The paper’s senior author, MIT professor Pattie Maes, explained:
“We basically can’t live without our cellphones, our digital devices. But at the moment, the use of those devices is very disruptive…. My students and I have for a very long time been experimenting with new form factors and new types of experience that enable people to still benefit from all the wonderful knowledge and services that these devices give us, but do it in a way that lets them remain in the present.”