An international team of researchers has developed a “smart throat” system that enables stroke survivors and patients suffering from dysarthria to regain natural speech. This is detailed in the published study.
The wearable device employs special sensors and artificial intelligence to process speech and emotional signals in real time.
The installed sensors analyze the vibration of throat muscles and the pulse on the carotid artery, while a large language model is responsible for speech processing. It analyzes speech and emotional signals, refining and expanding sentences to match the intended meaning of the user.
The “smart throat” is claimed to convert silent speech into coherent sentences without delays.
During tests on five patients with dysarthria, the device’s word error rate was 4.2% and 2.9% for sentences.
“The system generates personalized, contextually appropriate sentences that accurately reflect the meaning intended by the patients,” the study states.
The design of the wearable gadget resembles a choker. It incorporates highly sensitive sensors and a wireless communication module for continuous data transmission with minimal energy consumption, ensuring uninterrupted operation throughout the day.
The research suggests broader applications for the “smart throat,” including support for other neurological conditions like Parkinson’s disease and multilingual adaptation.
The team is currently focused on enhancing the device’s user-friendliness.
Back in July, researchers from the University of Florida developed an AI system that effectively tracks the progression of Parkinson’s disease through patients’ hand movements.