Dave asks the question, "I know you don't agree with using computer software to clean EVP so can you tell us how you analyze EVP and can we do it without spending a fortune in equipment?"
Certainly. And most EVP analysis requires nothing more than an open mind and a bit of critical thinking to resolve.
A few years ago orbs were considered paranormal. But now, even the most ardent believers have to admit that dust can cause orbs in pictures. We can prove dust is responsible; anyone can do the experiment and observe for themselves. (That is another topic, see "Orbs Explained" under photography. ) But EVP has yet to be given the same level of analysis. Many people still bring in recordings that are questionable at best claiming they are proof of spirit communication. Those of us who do analysis use methods which serve to debunk most EVP; perhaps if others applied the same critical analysis instead of just assuming something is communicating we can improve the quality of evidence available for study. This essay will outline the basic methods I use when I receive an EVP recording.
To begin with, any evidence requires a summary of how it was obtained. This includes the make, model, and the settings of any equipment used in obtaining it. One should also provide any environmental records that may have influenced the recording process. While these may not directly influence what is on the recording, they could sway one's opinion of subjective matter. For that reason alone they become a part of the evidence process. And of course knowing if anyone was present in the room at the time the recording was made is helpful in eliminating that source of contamination.
Let's take a hypothetical EVP through the process, step by step. that I would do. Of course this method may be altered if the evidence warrants, this is a basic outline only. The first thing is to rule out the most common form of false positive, pareidolia. This is done using the Rule of Three.
~ RULE OF THREE ~
Three people who are unfamiliar with the EVP listen to it individually two times only. Without consultation, they simply write down what they heard. If all three agree on its content, that is considered strong evidence pareidolia is not responsible. If two of the three hear the same thing, and the third substantially agrees, it is also evidence of not pareidolia although not nearly as strong.
However if only two of the three agree it is an indication the recording is probably not valid. Unless some overwhelming supporting evidence is available it should be discounted as pareidolia.
And of course if none agree it is an indication of pareidolia and is discounted.
Consider, if someone speaks directly to a group of people there is usually little discrepancy in what was said. But if the message is garbled it becomes somewhat subjective in content. The more garbled it becomes, the more subjective. Eventually it reaches a point where no one can understand it and subjectivity / pareidolia takes over. The Rule Of three simply establishes the point where this transition takes place. As such it also determines when an EVP no longer remains credible.
But for sake of discussion we will assume this EVP passes the Rule Of Three review. The next step is to determine its source; background sounds, recording artifact, or something else. For that we need the buffer area before and after the EVP. I always want a minute or so of recording before and after the EVP along with the actual EVP. Any physical sounds in the room will leave a certain ambiance related to the room. This is compared to that which accompanies the EVP. In the process you may hear something which matches phonemes related to the EVP.
All electronics introduces certain sounds. The background hiss is one example. This becomes a subtle part of any recording we must deal with. And we can't forget about the location itself. Buildings creak, an animal may call out in the distance, traffic may go by outside, including someone with a loud radio playing. All of these contribute their own unique sounds to the background.
Then there are the Sounds like that of a freezer for example may mimic a pulsating sound of a "P" for example. Alone it is clearly a freezer but if some other sound occurs at the same time two may blend into something like a word. This can be confusing and lead one to think they caught something when in fact the environment was responsible.
Or listen for the sound of traffic outside. If a car drives past you now have that sound to compare with anything else on the recording as a reference. Which also may be something to consider if you hear voices just as the car passes. Might the driver had that radio blaring?
So listen to each sound individually. Note the ambiance of each. Does one sound different than the other? Does part of a word sound like it was close by while the next phoneme sound distant? That is a clear indication the word was not formed as an EVP; rather a combination of two sounds occurring in succession. Remember, the portion of the recording before and after the EVP is just as important as the EVP itself.
What about the content of the EVP itself? What if it answered a question I asked? That is really not always significant unless you asked the right question. If the investigator asked a "Yes-No" question, the EVP answering "Yes" or "No" is no more evidence than saying "Knock once for yes, twice for no", and getting the sound of a rafter creaking. The single word answer is meaningless.
Why? Go back to Step 2 and consider all the sounds you may pick up. Then consider that much speech can be done using only 25 phonemes. How hard would it be for two or three sounds to fall together in a manner to form a simple, short word? Also consider, there may be no correct answer. So even if you got a "Yes" or "no" would you know which is correct?
The solution here is length of the message. If you ask questions that require some degree of response, several words for example, and get a sentence in response we can rule out many random sounds that might occur while the recording was being made. Rafters creak once, maybe twice in succession, not 15 or 20 times.
Also, those short responses are much easier duplicated by background sounds. Consider the electronics hiss mentioned earlier. You already have 1/3 of the phonemes required to form the word "Yes" present. Anything making a rubbing sound can sound like an "eeh" phoneme. Put the "eeh" in front of the "SS" electronic hiss and you have an "es" sound. Now ask a question requiring a "yes-No" answer and you get "es". Audio pareidolia will supply the missing "Y" sound and you have the making of a false positive.
To summarize these three steps we see the importance of minimizing extraneous noise since that leads to pareidolia and has nothing to do with "helping" any spirit communicate. It also shoots a hole in the common belief that one should create background sound such as white noise or other contamination. Sure, it helps get more recordings but they are also false positives.
This also points out the importance of using quality recording equipment. (That is covered elsewhere!) Cheap voice recorders by nature create artifacts which manifest themselves as noise. This also contributes to false positives, thus the reasoning some claim that using cheap recorders gets more EVP.
Assuming diligence was used in obtaining your EVP, and quality equipment in recording it, using the above methods will eliminate a large part of what you do capture. And it will also leave fewer, but much more credible, recordings that might be EVP.
Dave also asks a follow up question, "If this is all one needs, what do you gain by using your elaborate analysis methods such as spectrum analyzers and oscilloscopes? "
Once you get through the initial steps you will find that about 5% of your recordings seem to be valid. Now this doesn't make them so, but you certainly want to test them to the fullest extent possible to make sure. There is no short cut or cheap way past this; if you get an EVP to this level and you lack the equipment to do a complete analysis you may want to submit it to someone who can put through the tests. But I will outline a bit of what is done so you can get an idea of the methodology applied.
This is the most complex area and I can only touch on a few points here. It would require a book to cover all the methods and what may be done to enhance and further define an EVP. But for sake of discussion we will assume your recording requires in depth analysis. It has already passed the first three criteria outlined above. Now I am going to dig deeper into it.
The first thing will be to run it through a spectrum analyzer. This will display all the frequencies present in the recording. Some may be voice, others may be extraneous noise. The spectrum analyzer will show all, and a comparison against the real time recording allows the source to be identified.
The next step requires the use of narrow band filtering. The EVP is played through the filters, then, using the frequencies obtained from the spectrum analyzer, one by one these are omitted then restored while both observing the EVP on an oscilloscope and listening to it. The EVP may be called into question if its content changes when a frequency related to background noise is removed. That raises the possibility that the noise contributed to the recording, not actual communication.
Of course the filtering may reveal nothing, the recording may not have any matches between the background frequencies and voice patterns. The next step is to observe individual phonemes on the oscilloscope. These are compared to known speech patterns. For example, a spoken "sss" sound has a certain wave form that differs from the "SSS" created by electronic noise. The reason is the source of each. The spoken sound is modeled by the human vocal tract (notably the tongue and teeth in this case while the electronic sound is created by the flow of electrons through a circuit. These create radically different patterns although the sounds may be similar.
Earlier the sound of a "P" was mentioned as possibly coming from the compressor in a refrigerator or freezer. If such a sound is displayed on a scope, its nature is very apparent. Motors and such make a sort of rumble that has a slow rise time. The "P" sound made by the mouth is much faster. This rise time is easily seen on a scope at the start of the phoneme in question.
Each phoneme of speech has its own unique characteristics. And each must be taken apart and examined one at a time. Sometimes this is aided by using bandpass filtering; other times it's not needed. The point is this becomes a time consuming process, and no single method can be set. Each step along the way is determined by earlier tests on the recording. Some phonemes may match speech patterns, others may not be as definitive. Sometimes a single evaluation reveals the truth, other times several tests and attempts may be needed to determine the nature of a particular sound. In the end you will have to evaluate the recording noting how many match versus those that don't. One or two mismatches generally doesn't mean much. But if half or more don't follow speech patterns doubts begin to surface as to the validity of the recording. Too many doubts means you may have to discount the recording after all.