The Audio Test for Potential Coronavirus Infection Built by Voice Tech Startups Has a New Home
The audio test for potential COVID-19 infection covered by Voicebot in an exclusive two weeks ago is now available to the public. The COVID Voice Detector test, created in a partnership between enterprise voice assistant developer Voca.ai and Carnegie Mellon University, analyzes the probability of a user’s infection by examining the sound of their voice and cough.
The test starts with some basic questions about your health and any previous exposure or infection by the coronavirus before diving into the audio section. Users record the sound of their cough, count to 20, recite the alphabet, and measure how long they can make certain vowel sounds. All of that information is then measured and scored for lung health and potential COVID-19 infection. The user is encouraged to complete the test regularly, not only to see if their health has changed but because all of the information is connected to a dataset that researchers can use to create and improve tests for infection.
“In viruses like the coronavirus that harm the respiratory system, there’s a high probability we might find a pattern in the way a person speaks using voice biomarkers research,” Voca.ai co-founder Alan Bekker previously told Voicebot.
Between 20,000 and 30,000 people had already taken the test on Voca.ai’s website two weeks ago. With the official COVID Voice Detector website up and connected to Carnegie Mellon University’s website, that number is likely to grow dramatically. Voca.ai and CMU are also working with data analysis and healthcare firms telling.ai, hat-ai.com, and respiredx.com, as well as Amazon Web Services, to collate and understand the data collected. Bekker said the plan is to offer the dataset openly for other researchers to use and improve upon once there’s enough to share.
More Vocal Tests
This isn’t the only venture in measuring coughing sounds to predict coronavirus infection. Indian startup Salcit Technologies debuted an app just this week to do so named kAs, the Sanskrit word for cough. kAs uses the questionnaire created by the World Health Organization for checking the probability of infection before asking users to cough into their smartphone microphone. The cough is analyzed by artificial intelligence platform Swaasa, rendering a risk score out of 10. Salcit has a patent for analyzing risk based on cough sounds and is in discussion with the Indian health department, possibly augmenting India’s coronavirus WhatsApp chatbot with the test.
Determining the probability of infection before getting tested is vital as healthcare systems strain to handle the load of the ongoing pandemic. It’s why governments and healthcare organizations are leaning on AI developed by companies like Orbita and Hyro. Triage tests that can stop those who don’t need to go to the hospital from doing so could save a lot of lives and keep those who are unlikely to be infected from getting exposed.