EVP investigation techniques - what actually works and what's nonsense

I've been doing EVP (Electronic Voice Phenomenon) work for about 6 years now and I've learned a lot about what actually produces useful results versus what's just confirmation bias and audio pareidolia. Thought I'd share some honest reflections.

What probably doesn't work: Using your mobile phone's voice memo app. The audio quality is just too poor and noise floor too high - you'll hear 'messages' in the white noise whether they're there or not. Similarly, those cheap digital recorders from Poundland? Same problem. You need decent equipment because garbage in = garbage out.

What might actually work: High-quality digital recorders (Zoom H6, Roland R-05, that level of equipment) in locations with controlled acoustics. I've had some genuinely interesting recordings in old buildings, particularly around areas with documented historical events. Nothing I'd call 'proof of afterlife', but definitely unexplained sounds that don't fit normal categories.

The honest bit: I've also recorded thousands of hours of absolute silence and noise. For every one interesting recording, there's 100 hours of nothing. That's not exciting, but it's real. I've also definitely captured things that turned out to have mundane explanations - pipes, animals, neighbouring properties, structural sounds.

The filtering issue: This is the big one. If I record 40 hours in a location and get one weird voice-like sound, how much filtering can I do before I'm essentially manufacturing the result? That's the ethical line I try to stay on the right side of.

Curious what other EVP investigators have found. Are you getting results? How do you handle the question of audio pareidolia?

Honest answer: I've done about 150 EVP sessions over 5 years and I've got maybe 5 recordings that I genuinely can't explain. But those 5 don't prove ghosts exist - they prove that sometimes recording equipment picks up sounds we can't categorise. Might be paranormal, might be environmental, might be something we haven't thought of.

The pareidolia issue is real though. Audio pareidolia works the same way as seeing faces in clouds - your brain pattern-matches to human speech because that's what it's designed to do. I try to use blind analysis where possible - get recordings reviewed by people who don't know where they were recorded or what I expected to find.

For every one interesting recording, there's 100 hours of nothing

The equipment point is crucial. I was getting 'results' on cheap recorders for ages, thought I was proper ghosthunter. Upgraded to professional-grade kit and suddenly my 'results' dropped about 80% because I could hear what was actually just noise. Bit depressing but also honestly clarifying.

Have you looked into the frequency-following response theory? Some researchers claim EVP might be our brains interpreting random white noise as speech at a subconscious level due to specific frequency ranges. Would explain why it's so difficult to verify - not because ghosts aren't trying to communicate, but because we're essentially mind-reading white noise.