Health AI Part 3: Corona Detection from the Voice
©2021 audEERING GmbH Susanne Reichert 14.05.2021
Many diseases can be detected from the voice as they affect speech production. This also includes COVID-19 as it is a respiratory disease. In this third part of our Health AI series, we will explain how the disease can be detected from the voice.
Voice biomarkers for COVID-19
Scientific studies have shown that coughing, sneezing, and other changes in voice are reliable biomarkers for the detection of COVID-19. That means, that the disease affects the respiratory system. With a lot of relevant data from sick as well as healthy persons, an AI model can be trained to distinguish if a person is infected with the virus or not.
Our portal AI SoundLab for gathering COVID-19 voice data
That is the reason, why we have developed our portal AI Soundlab: to gather relevant data for training our AI to detect COVID-19. In the study, we collect data like for example speech samples of a person reading a text, coughing sounds and breathing sounds from inhaling and exhaling.
Furthermore, the platform enables us to monitor symptoms over a longer period of time. If someone gets tested positively with a PCR or rapid antigen test, we ask them to do the study again to see how their voice is affected.
Advantages of smartphone voice test
Based on the data we have gathered with our platform AI SoundLab, we will then create an app, which can enable any smartphone user to make short voice recordings and get an indication, if you have COVID-19. In case of a high probability to be positive, the users should verify their result e.g. with a PCR test.
Our vision is to make a normal life possible again with this simple but powerful solution. Help us to turn this vision into reality and record data on our website: aisoundlab.audeering.com