Digital avatar, AI tech enable paralysed woman to speak

Loss of ability to speak or express one's thoughts can be suffocating. Add to it serious impairment like paralysis which caused it in the first place, you get a situation you'd not wish on anybody. But unfortunately, there are many who suffer this fate.

In what can be a major help to such patients and their loved ones, researchers at the University of California have developed a technology with the help of which a paralysed personwithout speech can 'speak'.

The researchers have facilitated help from a digital avatar which is capable of not only delivering speech but also indicating facial expressions such as smiling, frowning or surprise.

But how does the avatar know what to speak?

The avatar gets signals from a Brain-Computer Interface (BCI)

The technology involves the use of tiny electrodes implanted on the surface of the patient's brain. These electrodes detect electrical activity from the part of the brain that governs speech and facial movements. These signals are then translated into the avatar's speech and facial expressions.

A crude explanation of the technology would bethat the avatar 'gets' what the paralysed person wants to say by way of what he 'thinks'.

“Our goal is to restore a full, embodied way of communicating, which is really the most natural way for us to talk with others,” said Prof Edward Chang, who led the work at the University of California, San Francisco (UCSF). “These advancements bring us much closer to making this a real solution for patients.”

He was quoted by The Guardian.

The patient in this case is Ann (47). She reportedly suffered a brainstem stroke and has been severely paralysed for more than 18 years.

Researchers implanted a small, paper-thin sheet of 253 electrodes on the surface of Ann's brain. The implant was made over a region that was critical for speech. The electrodes caught brain signals which would normally have controlled her tongue, jaw, face and larynx.

Ann then worked with the team to train an AI algorithm in order to calibrate it with respect to her unique brain signals.

This enabled the computer to learn 39 distinct sounds. A language model similar to ChatGPT was then used to 'translate' Ann's brain signals into sentences.

WATCH WION LIVE HERE