A team of scientists at the University of Augsburg in Germany has developed a speech recognition app that can detect Covid-19 infections with high accuracy. Led by Björn Schuller, the team has been working on the app since March 2020, using voice recordings from Wuhan, China, as the initial data source. The app uses a neural network to compare speech patterns of people with and without Covid-19 infections, with a current accuracy rate of just over 80%. The team aims to make the app publicly available, but it will need more infected and non-infected speech samples to improve its accuracy.
The app has the potential to provide real-time data on the movement and spread of Covid-19 infections, according to Schuller. It could be used by both general practitioners and individuals who are unsure whether to seek medical attention or pay for a lab test. The app could also help to relieve the burden on overwhelmed laboratories and ensure that people who are likely to have Covid-19 are treated promptly. However, the app will need to be integrated into an official Covid-19 app or marketed as a general health analysis application, as Google Play only allows one official Covid-19 app per country.
The team is committed to ensuring the privacy and ethics of the app, and will continue to collect data to improve its accuracy before making it publicly available. The app has the potential to be a valuable tool in the fight against Covid-19, and its development is a testament to the power of technology in addressing global health challenges.
