Implant recipient demonstrates near-perfect communication with artificial intelligence-generated voice
Researchers at the University of California, Davis, have advanced a brain-computer interface (BCI) that translates brain signals into speech with an astounding 97% accuracy. As per a press release, this makes it the most accurate system to date in the field. A study on this innovation was published in the "New England Journal of Medicine".
In July 2023, the team implanted the device in 45-year-old Casey Harrell, who had been battling amyotrophic lateral sclerosis (ALS) for five years. This progressive disease deteriorates the nerve cells responsible for body movement control.
According to Aerzteblatt.de, Harrell's speaking speed had plummeted to 6.8 words per minute, whereas the typical rate for an English speaker is 160 words per minute. He relied on slowly typing words on a screen using a gyroscopic mouse that interpreted his head movements into a cursor.
Microelectrodes instead of nerve cells
Four arrays of microelectrodes were placed on Harrell's brain surface, each containing 256 wires that capture motor signals intended for the lips, jaw, tongue, and larynx muscles, which are typically transmitted via nerve cells. These signals are then forwarded to a computer, which translates them with software. This process happens almost in real-time, enabling the patient to speak fluently with occasional word pauses, as reported by Aerzteblatt.de.
The patient managed to utilize the system shortly after activation, the researchers affirmed. In the initial language data training session, the BCI achieved a 99.6% word accuracy with a 50-word vocabulary within 30 minutes.
"When we first tried the system, he cried with joy when the correct words appeared on the screen," said neuroscientist Sergey Stavisky. "Everyone present also cried." In the second session, the potential vocabulary expanded to 125,000 words, with the BCI achieving an accuracy of 90.2% after an additional 1.4 hours of training data.
Record-breaking precision
Stavisky's colleague David Brandman commented that their work showcases the most precise speech neuroprosthesis reported to date. "At this point, we can decipher what Casey is trying to say correctly 97% of the time. That's more accurate than many commercial smartphone apps that attempt to interpret a person's voice." This technology is revolutionary because it offers hope to people eager to speak but unable to, Brandman asserted. "I hope that technology like this speech BCI will aid future patients in communicating with their loved ones."
Although similar technology has allowed other patients to speak again through brain implants, previous speech BCIs often suffered from word errors, explained Brandman. "That made it challenging for the user to be consistently understood and posed a significant communication barrier. Our goal was to develop a system that allows the user to be understood whenever they want to speak."
Own voice through AI
A remarkable experience for Harrell is using his own voice. What he says is not merely displayed on the screen. Scientists trained an AI model for speech output through PC speakers using a speech he delivered prior to being diagnosed with ALS.
You can observe and listen to the effectiveness of this technique in a video from the university. Harrell confirmed that the voice matches perfectly. "It causes people to cry who haven't heard me in a long time." He hopes that soon everyone in his situation will have the opportunity to utilize such a device to aid communication.
Harrell revealed that he continues to work full-time for climate protection, and this device facilitates his work more effectively and efficiently. "It's so frustrating and demoralizing not to be able to communicate," Harrell acknowledged. "It's like being trapped." This technology will help those impacted return to life and society.
Following the advancements in Neurology, researchers have developed a groundbreaking brain-computer interface (BCI) that translates brain signals into speech with an exceptional 97% accuracy, making it the most accurate system in the field. This innovation could potentially revolutionize the lives of individuals battling diseases that affect speech, such as amyotrophic lateral sclerosis (ALS).
In the field of Neurology, Stavisky's team's work on this speech neuroprosthesis has demonstrated the most precise speech decoding reported to date. This technology allows individuals like Casey Harrell to communicate with a 97% accuracy, significantly reducing communication barriers.
