Audio signals from the body could potentially allow for scalable and affordable medical
diagnosis in various fields: this could be even more powerful if studied in combination with
wearable and mobile systems able to run models on device. In this talk I will fist introduce
the general concepts and then describe our large scale data collection of respiratory sounds
aimed at understanding if COVID-19 could be diagnosed through them effectively. Data can
be contributed and information found at covid-19-sounds.org.