Two versions of the same audio. Switch freely — playback keeps its position, so you are comparing two sounds rather than a sound and a memory. Pick whichever you prefer.
About ten minutes. Ten short trials. There is no right answer — some pairs are very close, and a few are deliberately identical to check the test itself is working.
Please use headphones or monitors if you have them, in a quiet spot. Laptop speakers are still useful data — just tell us, because "held up on monitors, vanished on laptops" is itself a result.