Researchers from Duke University have developed a groundbreaking speech prosthetic that can translate a person’s brain signals into spoken words. This new technology, recently published in the journal Nature Communications, has the potential to help individuals with neurological disorders, such as ALS or locked-in syndrome, who are unable to speak.
The current communication tools available for these patients are often slow and cumbersome. The speech decoding rate is only about 78 words per minute, while people typically speak around 150 words per minute. This lag is due to the limited number of brain activity sensors that can be placed on a thin piece of material on the surface of the brain.
To overcome this limitation, the researchers collaborated with biomedical engineering experts to create high-density, ultra-thin, and flexible brain sensors. They developed an implant that contains 256 microscopic brain sensors packed onto a postage stamp-sized piece of flexible plastic. These sensors allow for better distinction between signals from neighboring brain cells, leading to more accurate predictions of intended speech.
During the experiment, the researchers recruited four patients undergoing brain surgery for other conditions. These patients temporarily had the device implanted, and they participated in a listen-and-repeat activity. The device recorded the activity from the speech motor cortex as the patients spoke a series of nonsense words.
The recorded data was then fed into a machine learning algorithm, which accurately predicted the sounds based solely on the brain activity recordings. The decoder achieved an accuracy of 84% for the first sound in a string of three sounds that composed a nonsense word. However, accuracy dropped when the decoder processed sounds in the middle or at the end of a word, especially if the sounds were similar.
Overall, the decoder achieved an accuracy rate of 40%, an impressive result considering it was working with only 90 seconds of spoken data. The researchers are now working on developing a wireless version of the device using a $2.4 million grant from the National Institutes of Health. This cordless version would provide more freedom of movement for users.
While this breakthrough is promising, there is still work to be done before the speech prosthetic is readily available. The researchers acknowledge that the current speed is slower than natural speech but believe that with further development, they can achieve higher speeds. The research was supported by grants from the National Institutes for Health, the Department of Defense, the Klingenstein-Simons Foundation, and the Duke Institute for Brain Sciences.
1. Source: Coherent Market Insights, Public sources, Desk research
2. We have leveraged AI tools to mine information and compile it