[ad_1]
An unprecedented brain-to-text system developed in Australia called DeWave allows users to turn their silent thoughts into written words, using only a wearable cap.
The technology, developed by researchers at the University of Technology Sydney (UTS), decoded brain signals captured by an electroencephalogram (EEG) into text. It represents a breakthrough in directly translating raw EEG data into language.
During testing, participants wearing the EEG masks silently read sentences while their brain activity was recorded. DeWave then processed these brainwaves to generate corresponding text, with an accuracy of more than 40%.
This surpasses previous EEG-based systems by 3% and is closer to the 90% accuracy of conventional speech and language translation software. The developers aim to ultimately achieve the same level of performance.
Other existing technologies that decode thoughts are highly impractical for everyday use. Invasive electrode implants and bulky MRI machines come with significant disadvantages. DeWave’s simple cap interface offers an easy-to-use alternative using AI and neural decoding.
The key innovation enabled by DeWave is the use of discrete coding to convert EEG signals into a symbolic code. This code is then linked to text using a ‘codebook’ that points to specific words.
Major language models such as BERT and GPT help relate encoded brain activity patterns to likely word sequences. Extensive training in reading datasets made it possible to translate verbs with the highest accuracy, although nouns proved to be more challenging.
Testing on a sizable sample of subjects strengthened DeWave’s reliability compared to previous technologies that were only evaluated in very small groups. This demonstrates its potential for various real-world applications.
Its most immediate use could be to help patients with stroke or paralysis communicate by decoding their thoughts. The hands-free nature of DeWave also creates new possibilities for controlling external devices.
For example, bionic arm movements can be controlled purely by thought, making the mobility of amputees easier. Robots can also be controlled and operated seamlessly using just the mind, without physical effort.
Looking ahead, DeWave can transform the way people interact with technology, enabling effortless, thought-based control. By eliminating physical input, it could increase accessibility for many user groups.
Despite the promise, some limitations remain with the current cap-based system. Accuracy still lags behind special neural implants and MRI brain imaging. There are also challenges in decoding exact nouns versus similar words.
Researchers emphasize that translating thoughts directly from the brain remains extremely complex. But leveraging innovations in AI, such as large language models, can help bridge the gap between neural signals and understandable text.
While further improvement is needed, DeWave represents a paradigm shift in brain-computer interfaces. Its portable nature and silent operation open up new potential applications for mental communication and interaction.
With this breakthrough, the futuristic vision of controlling the world through thought alone has come one step closer. The Australian makers of DeWave have paved the way for the next generation of fluid interfaces between minds and machines.