AI and EEG Transform Silent Thoughts to Text

Researchers at the GrapheneX-UTS Human-centric AI Centre, UTS, have unveiled a groundbreaking portable system decoding silent thoughts into text, aiding communication for those unable to speak due to conditions like stroke or paralysis. This non-invasive technology opens avenues for seamless human-machine communication, facilitating tasks such as operating bionic arms or robots.

Unlike previous methods requiring surgery or large MRI machines, their approach involves participants silently reading text passages while wearing an EEG cap. The cap records brain activity, and an AI model named DeWave translates EEG signals into language, representing a significant breakthrough in neural decoding. The study, spotlighted at the NeurIPS conference, demonstrated the technology’s potential with 29 participants, promising more adaptability than previous one or two-person tests. While achieving state-of-the-art performance, the model currently yields a 40% accuracy on BLEU-1, with hopes to reach levels comparable to traditional language translation or speech recognition programs. This research builds on UTS’s brain-computer interface technology, which commanded a quadruped robot using brainwaves in collaboration with the Australian Defence Force.


Source NeuroScienceNews

Author: Neurologica