An EPFL start-up is studying the possibility of identifying a person infected with the coronavirus using a non-invasive method based on voice analysis.
The smear in the nose, the days of isolation awaiting the verdict. Queues in front of hospitals and ephemeral screening counters deployed in response to waves of contamination.
What if all these measures could be avoided, simplified? Many scientists are struggling to find new, rapid, non-invasive methods to aid diagnosis.
A singular voice print
In the batch, an EPFL start-up, Vocalis Health stands out for its original and promising approach. She is studying the possibility of identifying the presence of the virus in a patient using voice analysis.
As early as March, at the peak of the first wave, the young company asked volunteers who had already tested positive for Covid-19 to record their voices in an application, by describing an image, then by counting from 50 to 70.
An artificial intelligence program isolated a voice print present in the voices of the sick that did not appear in that of the healthy. The start-up has also developed an application to detect the virus in the sound of a cough. Internet users wishing to participate in research can record their cough and send it to researchers via the site Coughvid.epfl.ch.
In the study phase, if the research bears fruit, it could one day allow everyone to know at any time and instantly whether or not they have the virus, by means of a voice recording on their smartphone. What effectively limit the risks of propagation.