Signal to noise ratio is a major consideration when analizing audio. Are you really hearing a voice or is it actually the lower frequencies of another sound that resembles a voice? One simple way to see how SNR affects audio is to play around with it in the software you are using for analysis and examine how the voice chganges as the signal to noise ratio is tweaked.
Below i have included five audio samples that contain the phrase "Can you hear the EVP?" They start out simple, with a decent signal strength but they get progressively harder to make out as the signal becomes weaker and embedded deeper into the noise floor. Can you find the phrase spoken in each file?
I'll follow up on this post in a week or so. Feel free to post your comments and findings. Test your might!