Being sick, whether with a cold or covid-19, changes your voice in a variety of ways. Now MIT Lincoln Laboratory researchers have found that processing speech signals may reveal such changes in covid-19 patients before the effects can be heard, potentially offering a way to identify asymptomatic cases.
Thomas Quatieri of the laboratory’s Human Health and Performance Systems Group had previously used signal processing to detect indicators of disease in the speech of people with neurological disorders such as ALS and Parkinson’s. Thinking about the symptoms of covid-19, he and his colleagues reasoned that inflammation and breathing difficulties would be likely to affect the loudness, pitch, steadiness, and resonance of patients’ voices, perhaps before obvious symptoms developed.
So they combed YouTube for interviews celebrities who’d tested positive for covid-19 had given while presumed asymptomatic. Then they downloaded older interviews with the same people and used algorithms to extract vocal features from each audio sample. The results suggest that the disease’s impact on muscle movement in the respiratory tract, the larynx, and articulatory features like the tongue, lips, and jaw causes subtle voice changes.
The team is now working to validate the data and, with Satra Ghosh at the McGovern Institute, exploring ways to use it in mobile apps for screening. They also hope to consider neurophysiological impacts linked to covid-19, like the loss of taste and smell, Quatieri says: “Those symptoms can affect speaking too.”
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
We can’t afford to stop solar geoengineering research
It is the wrong time to take this strategy for combating climate change off the table.
Meet Altos Labs, Silicon Valley’s latest wild bet on living forever
Funders of a deep-pocketed new "rejuvenation" startup are said to include Jeff Bezos and Yuri Milner.
The new version of GPT-3 is much better behaved (and should be less toxic)
OpenAI has trained its flagship language model to follow instructions, making it spit out less unwanted text—but there's still a way to go.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.