Mate, I'm strictly a UFO/lights-in-the-sky person so EVP isn't really my lane, but I've had something adjacent happen with my Zoom H5 recorder out on Dartmoor - I'm out there listening for unusual atmospheric sounds to cross-reference with sighting reports, and what I captured versus what I heard standing there have been wildly different on more than one occasion.
Whether that's the recorder picking up frequencies outside human hearing range, audio compression doing something weird, or actual paranormal shenanigans - honestly couldn't tell you. My money's on the recorder just being better at its job than my ears are.
But the psychological angle is interesting isn't it? You expect to hear something, so maybe live you're filling in gaps your brain invents, then the cold hard recording strips all that away. Or the reverse - playback suggestion kicks in and suddenly you're hearing words in static that weren't ". There". Before.
Either way, sounds like a proper documentation nightmare. Are you running any kind of control recordings in the same location at different times to rule out environmental factors? Curious whether the differences are consistent or totally random - that'd tell you something at least.
What recorder are you actually using for this?