hot take: today's noise cancelling processing—whether via H1/H2 chips in airpods or apps like zoom or discord—is what make calls unintelligible. a moderate level of background noise is tolerable when it's a constant pattern, whereas noise cancelling often makes this pattern stochastic as it analyzes and responds in real time.

the superior mic is actually your MBP (or MBA) despite its lack of aggressive noise cancelling, unless you're in an exceedingly loud environment that can actually drown you out.

another way to reframe this is to consider noise in images instead of sound

if you have an image with a relatively consistent level of noise or artifacts (e.g. film grain), even if its distribution is gaussian and not truly even, you will perceive that texture as part of the image. it likely won't bother you and may even lend the image a certain visual quality.

whereas if that same image displays inconsistent or unexpected levels of noise or artifacts (e.g. a large hair or a light leak), you will perceive it as an intrusion or distraction that draws your eye away from the content of the image.