There are some known exceptions, such as musicians, who can differentiate quite well. Which lets us assume, we can better this by training hearing properly.
For this music seems to be a key, since you can try to focus on specific parts like single instrument and try to ignore others, even though they are all present. This is done by the calculative power of our brain and happens in the signal processing stage of it. Not a hardware problem. It even uses additional information.
During the pandemic lots of people wore masks and I had trouble understanding many of them well. This isn't mainly a SNR problem, for me to get good differeniation seeing the persons mouth helps.
Sometimes I don't understand what people say, they repeat it and repeat it until my brain moves to the right place and everything is immediately clear.
I enjoy the occasional industrial death metal concert, here I noticed even with earplugs, shutting you ears with your hands, you can very well hear what a friend yells directly at you. Construction worker with earmuffs using insanely loud equipment, still exchange information by yelling. So even if an insane amount of noise is present and then damped to safe levels, direct communication is viable if SNR is good enough.
The problem is the processing stage, which requires additional training.
At least that is my theory. I assume loudness in public is at appropriate levels otherwise people would complain.