In a groundbreaking development, Australian researchers have created a non-invasive AI system that can translate silent thoughts into text. The technology, known as DeWave, uses electroencephalogram (EEG) recordings of brain waves to decode thoughts into written text. The researchers believe that with further improvements, DeWave could have applications in helping stroke and paralysis patients communicate, as well as making it easier for individuals to control machines like bionic arms or robots. Although DeWave currently achieves only 40% accuracy, the researchers hope to improve this to around 90%, which would bring it in line with conventional methods of language translation or speech recognition software. Unlike other methods of brain-to-text translation, DeWave does not require invasive surgery or bulky MRI machines, making it more practical for everyday use. The technology uses a combination of trained language models, including BERT and GPT, to match brain wave patterns with words and create meaningful sentences. Verbs were found to be translated most accurately, while nouns tended to be translated as groups of words with similar meanings. The study used a relatively large sample size of participants to account for variations in EEG wave distributions, making the research more reliable than previous studies.
