This advancement is raising hopes that brain-computer interfaces (BCIs) could be on the verge of transforming the lives of individuals who have lost their ability to speak due to conditions such as paralysis and amyotrophic lateral sclerosis (ALS).
Previously, patients had to rely on spelling out words using eye-tracking or small facial movements, which made natural speech impossible and required them to use slow speech synthesizers.
The latest technology uses small electrodes placed on the surface of the brain to detect electrical activity in the part of the brain that controls speech and facial movements.
These signals are directly converted into the speech of a digital avatar and facial expressions like smiling, frowning, or looking surprised.
Professor Edward Chang, who led the study at the University of California, San Francisco (UCSF), said, "Our goal is to recreate a truly natural form of communication that is as full and concrete as speaking with others. These developments are bringing us closer to making this a real solution for patients."
One patient, a 47-year-old woman named Ann, has been severely paralyzed since suffering a brainstem stroke over 18 years ago. She cannot speak or write and communicates using motion-tracking technology that allows her to select letters slowly, enabling her to form sentences of up to 14 words per minute.
She hopes the avatar technology will eventually allow her to work as a consultant.
The team placed a paper-thin rectangle with 253 electrodes into Ann's brain, targeting the critical area for speech.
After the implantation, Ann repeated various sentences to train the system's artificial intelligence algorithm.
The computer learned 39 different sounds and used a ChatGPT-style language model to translate signals into understandable sentences.
This was then used to control a personalized avatar that resembled Ann's pre-injury voice, based on a recording of a speech she made at her wedding.
