Universities of Cambridge and Beihang have developed a wearable device for people who lost clear speech after a stroke. An elastic collar with microsensors captures movements of the throat muscles and the pulsation of the carotid artery, after which an artificial intelligence system converts the signals into synthesized speech. Unlike similar neural implants, the device does not require surgery and is suitable for everyday use.
The technology’s main feature is its ability to recognize not only words, but also their emotional tone. The system distinguishes neutrality, relief, and irritation, and can also automatically expand short phrases into natural sentences if desired. Switching between modes is done by moving the head. The synthesis delay is about one second, and the speech sounds smooth and natural. User satisfaction was 55% higher compared with standard communication tools.
The new technology could also be applicable in Parkinson’s disease and amyotrophic lateral sclerosis (ALS), where articulation is affected, and may become a new model of communication for people who have effectively been deprived of their voice.
