Meta's Noninvasive Brain–Computer Interface Brain2Qwerty Achieves 61% Accuracy
Meta has released the open-source Brain2Qwerty v2, a noninvasive interface that translates brain signals captured via EEG or MEG into text. The system reached an average word accuracy of 61 percent in initial evaluations. This development marks a technical step forward in efforts to decode language from neural activity without requiring surgical implants.
Covered by 1 source
- IInfoQ AI↗Anthony AlfordJul 14