Part of the issue with keyboard audio is that it's very "noisy". It's like comparing two instances of white-ish noise. Statistics would be able to discern the instances immediately, but a human probably wouldn't.
>>snet0+(OP)
Another part of the issue is if the laptop has two microphones, it can distinguish a place for low-freq sounds. The human head cannot locate low frequency sound sources such as a sub-woofer in a 2.1 system.