Laboratory Notebook 8 October 2002
Voice Frequencies in a Spike
Spikes are a common form of electromagnetic phenomenon - there is nothing paranormal about spikes. A spike could come from a thunderstorm 30 miles away, from your neighbour's cooker switching itself off and on, from you pressing a key on your keyboard - and from making selections from a stream of white, brown or pink noise - to pick a few examples.
Normally spikes are largely ignored - but in an aircraft or a space-vehicle they are bad news. In the aerospace business, expert consultants are hired at vast expense to get rid of spikes or their effects. Spikes occur everywhere. The exact position of one wire relative to another can determine whether a system is bugged by spikes or not.
Any spike can be decomposed into a vast array of component frequencies. Just as a house could be de-composed into an array of planks and bricks and sheets of glass, etc., so can a spike be de-composed into component frequencies. The components of a house have to relate to the size of the house as a whole and so there is a correlation between the sizes of the planks and the sheets of glass and the bricks etc.
There is also a correlation between the nature of a spike and its component frequencies. This is the opposite of white noise.
In the case of white noise however, (which is also composed of lots of frequencies) the correlation is non-existent - it is zero - white noise is a totally random process.
So - the difference between random noise and the components of a spike lies in the relatively tight correlation between frequencies in the case of the spike as compared with the case of white noise - you could say it was case of comparing random noise with non-random noise.
Spikes come in all sizes and "widths" (durations). Around a computer the usual spikes will have durations of a fraction of a millisecond or even a microsecond.
The range of frequencies that are the components of the pulse are tightly correlated to the spike duration. A brief, fast spike will have its components in the 100s of kHz or MHz.
There is a theoretically perfect shape for a spike that is called a Dirac impulse or function - but you won't find that in nature. And this is another factor affecting the make-up on the components of the spike.
Lets say that a spike has a "width" of 40 micro-seconds, then all its component parts will be correlated somewhat to that brief instant and will be away above the audio range.
However, if you stretch that spike - if you make it wider and wider - until perhaps it is 4 or 40 milliseconds long then what you will get is a group of correlated frequencies in the audio range.
Now if you do a statistical job on this - and this results in a summary which presents an overall view - then there may not be a great difference between the statistical result for speech and that for a stretched spike.
My method of presenting the objectivity of the speech contained in EVP - a method which was invented unknown to me by the Italian engineer Dr Renato Orso before my time - is simply to look at the waveforms of the speech contained in the EVP as presented on a storage scope or on a PC using software such as CoolEdit.
Now if you take the word "cat" and you look at the waveforms of someone saying "cat" on the screen these waveforms will have a specific form which corresponds exactly to the sequence of sounds C, A, and T. Like those smart people who can tell you what someone is saying just by reading their lips at a distance there are also people who can tell you what someone was saying by reading the waveforms on the scope or PC.
So, if your EVP shows the waveform patterns that correspond to C, A, and T then that is what the EVP is saying. You can check on this by using standard vowels from the American Phonetics society to compare the patterns with, or you can say them yourself. It takes long experience and skill to be able to read waveforms like this but doing a comparison job is easier.
There are of course little problems - like accents - but handling that comes with experience.
Here is a recent example of how I got caught out. I was writing my paper for the SPR and the only recent EVP utterance that I had to hand included the word 'Morris'. Now to illustrate the above technique I took the 'o' out of Morris and showed the waveform of that. Then I took a phonetically standard 'o' and displayed that to show that they were the same.
But woops - they were not!! Panic!!
It wasn't an 'o' sound - it was an 'uh' sound.
So what the voice was actually saying was "Murris" - it had an accent. I knew that what was being said was Morris - but it was actually being pronounced as "Murris". Of course mentally I had discounted the accent and put down the word that I had heard.
The word spoken however was "Murris" - and all you have to do to say that is to realise that an 'o' is a sound made with a fairly open mouth whereas 'uh' is said with an almost closed mouth. So, just say Morris but with a less open mouth and you will hear "Murris".