A small adhesive patch placed on the neck can produce words instead of the vocal cords. The device uses AI

A small adhesive patch simply placed on the neck manages to produce a few words instead of the vocal cords, according to a study published on Tuesday, an invention that is still far from being able to concretely solve speech problems, informs AFP, quoted by Agerpres.

The device recognized what people intended to say. PHOTO (illustrative) Shutterstock

“The patch, a small square of a few centimeters, detects certain muscle movements which are then converted into electrical signals and interpreted by an algorithm (…) able to recognize words and translate them into language”said a press release from Nature Communications, the publication where the study appeared.

The device, which uses artificial intelligence, relies on the movements of the muscles of the larynx, not on those of the vocal cords. So, one day, it could partially replace them.

The main objective would be to create a very simple patch to apply, without needing any surgical intervention, which would allow disabled patients to be able to use speech again.

Currently, it can only reproduce phrases that have already been stored in its memory

However, researchers warn that it will be years before such a device can help patients in real life. The device has been tested, for the moment, only on eight people who did not have speech disorders.

For now, the patch can only play phrases that are already stored in its memory. In the study, it was about expressions like “Merry Christmas'' or “Ti love her''.

The device recognized what people intended to say

constraint “limit uses of our device”', Ziyuan Che, a researcher at the University of California in Los Angeles (UCLA) and the main author of the study, admitted to AFP. But the research, in which the device recognized what people intended to say in 95% of cases, is promising, the researcher assured.

Ziyuan Che says that more elaborate algorithms could one day allow the patch to produce expressions “without the need to pre-record voice signals''.