Brain impulses will be converted to speech.
The researchers trained an artificial intelligence to interpret the data of brain activity and convert them to speech.
Three teams of scientists received data in the course of operations to remove brain tumors, as well as via electrodes attached to the brain of patients with epilepsy to identify the source of seizures.
The models had to train on each person separately, because the brain signals that are converted into speech, the individual. The accuracy and clarity of the generated speech varies from 40 to 80 percent. However, researchers have not yet figured out how to interpret the imaginary inner speech — that is, the signals, when someone is a mental dialogue with yourself.
One approach is to let the patient listen to speech, recreated by the computer, and adjust a person’s thoughts to produce the desired results. At the same time, it is possible to train the network to understand the way of thinking of the individual patient.
The study is in its early stages, but if successful, people with diseases accompanied by speech disorder (e.g. amyotrophic lateral sclerosis) will have an easier way of communication.