A new study coordinated by Sergey Stavisky and published on Nature tells how a 45 -year -old man suffering from ALS (Amyotrophic lateral sclerosis, a disease that afflicts control of the movement) has returned to speak Thanks to abrain-computer interface Super advanced call Braingate2. Have been implanted 256 electrodes in his brain for light the signs linked to movements of the mouth. THE’artificial intelligencecoached with vocal recordings of the man prior to his illness, has managed to reconstruct it surprisingly naturally, in a tone, rhythm and even hesitations. No longer a voice roboticsbut true communication, full of emotions. A huge step forward to restore expressiveness to those who can no longer speak.
Read the intentions of movement from the brain signals of the mouth
To allow artificial intelligence to elaborate the brain signalsthe patient underwent an insertion intervention of 256 silicone electrodes In a specific motor area of the cerebral cortex: the Little Ventral Girovery close and closely connected to the most famous verbal production area par excellence, the Broca area.
We are therefore talking about anKey area for programming and production of all those Oroofaccial movements (who concern mouth and face) who activate when we move our mouths to speak. Applying some Deep Learning algorithms Specially trained on the signals that come from this precise brain area, Wairagkar and its colleagues from the University of California have managed to have as a result of a flow of sounds starting from the neural processes that man produces in his brain, which concern the decoding of Fonemi, Toni, Speed and, in general, all those paralinguistic components present in the spoken speech.
The novelty compared to the past
Similar machine interfaces similar are already be tested more or less effectively in the past. A less recent article, always on Nature, dated April 2022, informed us of how, already three years agoexisted At least 35 people With a similar system positioned on the long -term cerebral cortex, for the most disparate problems: connection with robotic prostheses, communication with synthetic voice or written communication on screens through imagination in people with paralysis.
The step forward which has been done in these days concerns a New level of naturalness Unlocked thanks to the use of the most sophisticated artificial intelligence. Do you think that, in the case in question, the IA has been trained on the vocal traces of the man prior to his illness, so as to be able to reproduce the exact frequencies of his voice. A voice no longer flat and robotic, but rich in nuancesuncertainties, neologisms, changing tones, emphasis and suspensions. Everything that makes one “human” voice and that allows us to understand the intentions of the speaker, the emotions who are underneath, and then enter more in empathy. With this goal these qualities can be returned to people with motor diseases that hinder expressiveness.
