Australian Scientists Use AI to Translate Brainwaves Into Text, Offering Hope for Speech Disorders

Australian Researchers Develop AI Model That Turns Brainwaves Into Words

Scientists at the University of Technology Sydney (UTS) have created an AI system that decodes brainwaves captured via electroencephalogram (EEG) into readable text. The AI uses deep learning to translate EEG signals into specific words and phrases, effectively converting thought patterns into language.

The model was developed by PhD student Charles (Jinzhao) Zhou along with supervisors Chin-Teng Lin and Dr Leong. When wearing a 128-electrode EEG cap, the AI successfully interpreted Dr Leong’s silent thoughts, producing the sentence, "I am jumping happily, it's just me." This demonstrates the system’s potential to recognize internal speech without any vocalization.

How the AI Model Works

The AI filters out noise and clarifies overlapping brain signals recorded on the skull's surface. Since EEG signals come from multiple brain areas simultaneously, this filtering is essential to isolate the signals that correspond to specific words.

Currently, the AI is trained on a limited set of words and sentences to simplify recognition. While this restricts its vocabulary, it lays important groundwork for more advanced decoding in the future.

Non-Invasive Approach and Its Challenges

Unlike invasive brain-machine interfaces such as Elon Musk's Neuralink, this system relies on non-invasive EEG. According to Chin-Teng Lin, this limits precision because the electrodes cannot be placed directly over brain regions responsible for language processing.

Despite this limitation, the technology holds promise for several practical applications:

• Stroke rehabilitation
• Speech therapy for individuals with autism
• Restoring communication ability in paralysis patients

Global Advances in EEG and AI Research

Worldwide, combining EEG with AI is yielding promising results. For instance, researchers at Mass General Brigham recently developed an AI tool that predicts cognitive decline years before symptoms appear. By analyzing subtle changes in brain activity during sleep, the system flagged 85% of individuals who later experienced decline, achieving an overall accuracy of 77%.

These advances highlight how EEG and AI integration can contribute significantly to neuroscience and medical diagnostics.

For professionals interested in AI applications in brain signal processing and other fields, exploring specialized courses can provide valuable knowledge. Resources like Complete AI Training offer comprehensive learning paths for deep learning, neural networks, and biomedical AI.