US researchers develop brain implant that translates thoughts into speech with 74% accuracy

American researchers have achieved a significant breakthrough in brain-computer interface (BCI) technology by developing an implant that translates thoughts into speech or text with high accuracy, offering new hope for individuals with speech or mobility disorders.

The implant functions as a decoder of “inner speech”, achieving accuracy rates of up to 74% in converting thoughts into speech during trials with four volunteers suffering from severe paralysis. The invention aims to assist individuals with speech or mobility impairments.

Neuroscientist Benyamin Meschede-Krasa from Stanford explains, “If you just need to think about speech instead of trying to speak, it could be something easier and faster for the person”.

The implant

How the implant works

The device measures neural activity and identifies patterns related to phonemes – the basic units of speech. Using machine learning, these patterns are decoded and synthesised into words and sentences.

Researchers observed similarities between brain signals produced when volunteers attempted to speak and when they simply imagined words. With appropriate algorithms, the BCI can recognise up to 125,000 words solely from internal thought.

Stanford neuroscientist Frank Willett notes, “We found that we could decode these signals well enough to demonstrate the principle, although not as well as with speech attempts”.

Prospects and challenges

Although accuracy is not yet consistent, experts believe that with improved implants and mapping of more brain regions, the technology will improve significantly.

An additional concern is privacy protection, as there is risk of decoding internal thoughts. Researchers are testing a special “thought passcode” to activate and deactivate the function, with 98% success rates.

The system identified patterns linking words with brain signals through machine learning technology, demonstrating the potential for natural communication restoration in paralysed patients.

Willett concludes, “The future of BCIs is bright. This work gives hope that BCI speech systems will someday be able to make communication as natural and comfortable as a normal conversation”.

(information from CNN.gr)