US EditionStroke sufferers get their voice back thanks to artificial intelligence
The Revoice device achieved 96% word accuracy and offers a non-invasive solution to help half of stroke survivors with speech impairment communicate naturally.
- Published January 19, 2026, the University of Cambridge Department of Engineering unveiled the Revoice device that uses ultra-sensitive fabric sensors and AI to decode silently mouthed speech.
- About half of stroke survivors develop dysarthria, and `When people have dysarthria following a stroke...` said Professor Luigi Occhipinti, highlighting the need for better assistive devices.
- In a focused test the system correctly decoded 95.8% of words and 97.1% of sentences using two AI agents and an embedded lightweight LLM with one-second total delay.
- Participants reported a 55% rise in satisfaction, the team plans a planned clinical study in Cambridge this year, and Professor Luigi Occhipinti said `This is about giving people their independence back`, while the Stroke Association welcomed the research.
- Although promising, the results are preliminary and limited by sample size, as researchers call for larger, long-term studies and plan future upgrades with flexible circuit boards and multilingual features, funded by British Council, Haleon and EPSRC.
15 Articles
15 Articles
A Wearable 'Intelligent Throat' Gives Stroke Patients Their Voice Back
The device decodes words, picks up emotional cues from pulse patterns, and can expand fragmentary thoughts into complete expressions. The post A Wearable ‘Intelligent Throat’ Gives Stroke Patients Their Voice Back appeared first on StudyFinds.
Scientists at Cambridge University in the United Kingdom have developed a vestibular device that involves the brain and uses artificial intelligence to “return” the voice to patients with such serious brain vascular accident (AVC). The goal is to recover the ability of these individuals to communicate naturally and flows.
The findings open the door to future applications of the device in the treatment and assistance of people affected not only by a stroke, ...
Coverage Details
Bias Distribution
- 43% of the sources are Center, 43% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
