Researchers at the GrapheneX-UTS Human-centric Artificial Intelligence Centre at the University of Technology Sydney have developed a groundbreaking system that translates silent thoughts into text using AI and EEG technology. This portable and non-invasive system offers new communication possibilities for individuals with speech impairments caused by illnesses or injuries. Unlike previous methods that required invasive surgery or MRI scanning, this system utilizes a wearable EEG cap to record brain activity and an AI model named DeWave to decode the signals into language. The technology shows promise for enhancing human-machine interactions and assisting those who cannot speak, including potential applications in controlling devices like bionic arms or robots.
The study conducted by the researchers involved participants silently reading passages of text while wearing the EEG cap. The cap recorded electrical brain activity through the scalp using an electroencephalogram. The captured EEG waves were then processed by the AI model, DeWave, which translated them into words and sentences by learning from large quantities of EEG data. This innovative approach to neural decoding marks a significant breakthrough in translating raw EEG waves directly into language.
Unlike previous technologies that required surgery or MRI scanning, this new system is portable and non-invasive, making it suitable for daily use. It can be used with or without additional aids such as eye-tracking. The study, which involved 29 participants, promotes robust and adaptable decoding technology since EEG waves differ between individuals.
Although the translation accuracy score is currently around 40% on BLEU-1, the researchers aim to improve it to a level comparable to traditional language translation or speech recognition programs, which typically achieves around 90% accuracy. The researchers believe that the model’s inclination towards synonymous pairs instead of precise translations for nouns is due to the brain’s processing of semantically similar words that produce similar brain wave patterns. However, despite these challenges, the model still yields meaningful results, aligning keywords and forming similar sentence structures.
The research builds upon previous brain-computer interface technology developed by the University of Technology Sydney in collaboration with the Australian Defence Force, which uses brainwaves to command a quadruped robot.
The groundbreaking technology developed by the researchers at the University of Technology Sydney has the potential to revolutionize communication for individuals with speech impairments and improve human-machine interactions. By translating silent thoughts into text using wearable EEG technology and AI, this portable and non-invasive system offers new communication possibilities and opens up opportunities for controlling devices like bionic arms or robots. With ongoing efforts to improve translation accuracy, this advancement has the potential to transform the lives of those who are unable to speak due to illness or injury.
