Shame we can’t burn the masks, they must be toxic
Better use:
What’s your problem with that/why is it funny?
I don’t know anything about the technology, but according to the article:
Users need to give information about their medical history, smoking status and demographics and record some respiratory sounds, such as coughing and reading a short sentence.
A voice analysis technique - called Mel-spectrogram - identified different voice features to “decompose the many properties of the participants’ voices”.
If it works, it works, I guess. You can do some clever stuff nowadays with automated analysis and machine learning (measuring someone’s heart rate using a cellphone camera, for example).
It stinks of scam.
Another way to get people or bed wetters tied to their ‘devices’.
How could it detect the difference between the flu, a cold, a hangover, a horse voice just on sentences, sounds and previous coughing?
Then again, no surprise, it is more accurate than the lateral flow test.
Yeah, good questions – I’ve no idea how it might be able to detect the difference between those things, especially other respiratory illnesses, irritation caused by pollution, etc. 
Though I’m not sure that anyone knows the “how” for some of this machine learning/AI stuff, including the designers. A lot of them are kind of black boxes where you input the training data (in this case voice recordings and COVID status, I guess) and it mysteriously processes them to find multidimensional correlations between some derived parameters (I suppose from different frequency regions) and the desired answer. The test is then the predictive accuracy of the model on some new data set (voice data without telling the model the COVID status).
I’m surprised, though not entirely shocked, if it works. But again, you can do some remarkably clever stuff with these kinds of models. I’m not sure that such an app is necessary…but that’s another story. Can’t really blame the researchers/developers for jumping on the bandwagon…
Or maybe it’s a scam like you say. Who knows in the post-truth era?
I imagine the technology would sort-of work - although as @cake said it’s unlikely to be specific to COVID, and surely there’s a subjoke there about the extreme degree to which everyone now thinks everything is caused by COVID. I think the main point was that it suggests the lateral flow tests are next-to-useless.
Why not both? If the tech works, you can give a diagnosis that’s at least as accurate as advertised, while also performing massive data harvesting for the purpose of some horrid scam down the line. Win-win! 
Don’t forget those that are completely and utterly “asymptomatic.” 
