Earlier quoted context omitted.
They did not test an infected person. They measured droplets, not virus. Disease progress is not relevant to this paper.
Ouch. I read droplets and thought 'virus'. That said, my point about the speaker not being representative is still correct. We have no idea if the speaker's speech is typical. It could be their flow is lower than the average.
They tested with four speakers and ran additional tests with one speaker. Results are consistent across the speakers. Consider also that they are measuring the relative filtration of different masks, not the absolute filtration.