Scientists at the University of California, San Francisco, have electrodes fed into volunteer brains and decoded signals in a part of the brain that is responsible for speech. These were then used in a computer simulation that could produce synthetic speech. The information was provided by Reuters.
According to scientists, speech is usually understandable. “When we first heard the results, we were in shock, we couldn't believe our ears. It was amazing how many aspects of real speech appeared in the synthesizer output, ”said co-author Josh Cartier.
Five volunteers who had already been introduced into the brain for electrodes were involved in the research. But none of them came to talk, so scientists want to work with silent people in the future.
Volunteers read the text out loud, while the instruments recorded brain activity. Thus, the researchers recognized the individual movements of the vocal tract needed to produce speech. For each of the volunteers, they created a virtual vocal tract that could only be controlled by brain activity to create artificial speech.
Nowadays, there are technologies that make it possible to "talk" who have lost their speech because of illness or disability. These devices track the movement of the eye or the movements of the facial muscles to form words. But this type of speech is very slow, creating no more than ten words per minute. Ordinary speech, however, flows at a rate of about 100 to 150 words per minute.
The newly developed procedure is designed to help people who have lost speech as a result of stroke, cerebral palsy, amyotrophic lateral sclerosis (ALS), Parkinson's disease, multiple sclerosis, brain injury or cancer.