Brain2Qwerty v2 enables real-time sentence decoding from non-invasive brain recordings, achieving 61% word accuracy and 78% for the best participant.

Brain2Qwerty v2 represents a significant advancement in the field of brain-computer interfaces, achieving a 61% word accuracy in real-time sentence decoding from non-invasive brain recordings. This new version builds on the previous iteration, v1, which introduced the concept of decoding brain activity into text without surgical implants.¹²³⁵

The system was trained on approximately 22,000 sentences from nine volunteer participants, each recorded for 10 hours using a magnetoencephalography (MEG) device while actively typing. The results indicate that Brain2Qwerty v2 can recover sentences coherently from noisy neural inputs, significantly improving upon the 8% word accuracy achieved by its predecessor.

Notably, the best participant achieved a remarkable 78% word accuracy, with more than half of all sentences decoded containing one word error or less. This level of performance approaches accuracy levels previously only attainable through invasive techniques that require brain surgery, marking a pivotal moment in non-invasive brain decoding technology.

As researchers continue to refine this technology, the implications for communication and assistive devices for individuals with speech impairments could be profound, potentially transforming how we interact with machines and each other.