Stanford develops thought-to-speech technology for paralysis

A new brain-computer interface developed at Stanford University may enable paralyzed individuals to communicate using only their thoughts. According to the research, the technology can decode inner speech with up to 74% accuracy and operates through a password-controlled security system.

The study, conducted at Stanford University in the United States, showed that brain-computer interfaces (BCIs) can decipher not only attempts to speak but also words spoken silently in the mind. The research was published in the journal Cell.

The technology is said to potentially help people with severe speech paralysis communicate solely through thought. Researcher Erin Kunz stated, "For the first time, we understood how the brain looks when you think about speaking."

MICROELECTRODES IMPLANTED IN THE BRAIN

In the study, microelectrodes were implanted in the motor cortex of four participants who were unable to speak due to ALS and brainstem paralysis. The participants either tried to say certain words or only thought them. Scientists decoded the brain signals using artificial intelligence.

UP TO 74% ACCURACY

The system was able to decode the sentences thought by the participants with up to 74% accuracy from a dictionary of 125,000 words. Researchers also developed a password-controlled security system to prevent unwanted inner speech from being read.

In experiments, the system only activated when the user thought of a specific phrase. The password "Chitty chitty bang bang" was recognized with over 98% accuracy.

Researcher Frank Willett said this technology could provide paralyzed individuals with a "communication experience close to natural speech" in the future.