A new device has been designed that can analyze the brainwaves of paralyzed patients and convert them into sentences on a computer screen in real time.
A "mind-reading" machine is able to decode brain activity as a person silently tries to spell words out phonetically to create complete sentences.
Experts say the neural prosthesis has the ability to restore communication to people who cannot speak or write because of paralysis.
Previous research showed that a similar system was able to decode up to 50 words.
However, this was limited to a specific vocabulary and the participant had to try to pronounce the words out loud, which required considerable effort, given their paralysis.
So Edward Chang and his colleagues at the University of California, Los Angeles, have designed a neural suit capable of translating brain activity into single letters to spell out entire sentences in real time.
Then they demonstrated its use in a person with limited contact due to severe paralysis of the voice and limbs.
The researchers extended the previous approach to a larger vocabulary by designing their system to decode brain activity associated with the phonemic alphabet.
In tests, the device was able to decode the volunteer's brain activity as they tried to silently speak each letter phonetically to produce sentences of 1,152 vocabulary words at a speed of 29.4 letters per minute, and an average letter error rate of 6.13 percent.
In other experiments, the researchers found that the approach was generalized to large vocabulary of more than 9,000 words, with an average error rate of 8.23%.
They say the results demonstrate the ability of silently controlled neural prosthetics to create sentences through a spelling-based approach using audio-coded words.
Among the successfully deciphered sentences are "Good morning", "You must be joking", "What do you mean", "I think this is very good" and "I will check".
But despite the device's success, the researchers caution that more work is needed to prove whether this approach can be used successfully in a larger number of participants.
"These findings demonstrate the clinical feasibility of silently-controlled speech-specific neural compensation to generate sentences from large vocabulary through a spelling-based approach, complementing previous demonstrations of direct whole-word decoding," the researchers wrote in their paper.