EVP - Classic Research

EVP - Classic Research
Posted by: Tony Bowmaker
Category: Sounds /

EVP has become a hot topic recently. However, it has a long history and ASSAP members were researching it long ago. Here is an article from 1991 by Tony Bowmaker. Many of the points he makes are still just as valid today, even though the technology has moved on from tape recorders to solid state memories.

From Anomaly Vol 9 Nov 1991

Some EVP research ends up in a field!

The following article is derived from a report on EVP written by Tony Bowmaker. The ideas presented are derived from his own research. The ideas may be of practical value to those wishing to pursue this line of research.


The electronic voice phenomenon (EVP) is an anomaly which manifests itself as a variety of sounds, some with human voicelike nuances. The effects are superimposed on the white noise or hissing sound generated by the active electronic components in equipment such as tape recorders and amplifiers. By employing a white noise generator, such as the microphone pre-amplifier built into a tape machine (if run with the mic/record level high) or a specially made noise producer using a diode or transistor as a source, one can record EVP events. A purpose-built generator is recommended because a reasonable level of noise can then be obtained. On playback, the voices and other sounds are quite difficult to hear at first, but sooner or later a strong event is obtained, where there is no doubt of its existence on repeated listening to the tape.

Precautions to Observe

To make EVP recordings scientifically acceptable and remove the possibility of false results, it was found necessary to consider the following procedures before starting. I have found it wise to erase the opposite side of a tape to the one in use before attempting to record. This is because any music or voice on the opposite side can sometimes leak over, though at a low level. It superimposes itself, in snatches, on the track that one is listening to, making one think results are coming along nicely. I have noticed this effect just once, before I started using the precaution above, and it sounded like a succession of quite convincing EVP sounds until some stranger voices gave the game away.

Another cause of false results can be local radio transmissions, like taxi cabs, causing breakthrough on the EVP equipment. Though it is obvious when this kind of interference occurs at high levels, it can sound convincing enough to people unfamiliar with it when at at lower levels. In my area, breakthrough involving taxis is quite common on tape machines and other audio equipment.

The Source of the Events

It seems that the white noise is part, parcel and source of the events and the tape machine stores the noise events. I say this because some texts on ‘Raudive voices’ give the impression that the voices are impressing themselves directly onto the magnetic tape itself, as if the noise signal was irrelevant. This could have been because earlier experimenters did not use external generators, but the noise generated by the equipment itself (initially how Jurgenson noticed it). One can illustrate this point by carefully monitoring the signal through the headphones in record monitor mode. This enables one to take note of the EVP events in real time and show correlations with the tape counter on playback. This is tricky, but it works most of the time, except for the odd event which was not evident in record monitor but appears as large as life on playback. This may be a good example of the subjective factors mentioned later.

How People Interpret the Sounds

I have found that, on most occasions when a member of a group listening to the same EVP event hears what he describes as a certain sound, the other members will more often have differing opinions on what was heard, other than agree that an event has occurred. Most people I have tried the sounds on hear what can be described as vocal nuances along with more abstract sounds. The non-voicelike sounds, ie. whistles, tones etc., do not bring any more mutual agreement. Never have I or any assistant listener interpreted any spoken word, contrary to some experiments. 

Hallucination and Suggestion As Factors

From my experience of listening with a group of people I tend to conclude that suggestion plays a part in the interpretation of EVP sounds. The effect seems to be more prominent at different times for different individuals listening to the same set of events. The subjective suggestion seems to add missing detail or embellishes an underlying event. I have noticed, for instance, that high frequency clarity or presence sometimes makes an event stand out, although it may not be audible to another listener. This becomes noticeable as an imaginary effect when the same event is repeated and found to lack the initial presence or clarity or even to have disappeared. Research so far suggests that for one’s imagination to build a subjective suggestion it must be based on the frequency content of the underlying basic event. The imagination would not build the same suggestion on unrelated frequencies or purely random events.

I will now give details of some subjective or hallucinatory effects which suggest that the imagination does not build on unrelated frequencies. Then I will describe an experiment which I carried out which suggests that imagination does not build on purely random events. The whole point is that the white noise which is used in EVP is derived from natural electron movement and is supposed to be one of the most random events known in nature, along with radioactivity. So even if the underlying events are more simple than is perceived they are still as anomalous, this fact increasing with the length of the event and its complexity. I will now describe the main subjective, or what I take to be hallucinatory, factors noticed in EVP work. These findings were produced with the help of an assistant listener.

  1. If a person expects a certain kind of sound, this will sometimes seem to manifest itself when the person is involved in listening. The sounds will sometimes appear to be there on repeated listening, then later on, say after a break, they will have either dropped in level, changed in quality or disappeared (rare).
  2. Prominent sounds in the neighbourhood, eg. a phone ringing or anything that one’s ears have been exposed to (though not necessarily aware of at the time) when taking part in EVP listening, can be incorporated. An example of this is, after exposure to the sound of a pulsing electronic alarm clock in another room and on another occasion to the noise of metal being struck outside, I found that these were manifesting themselves in certain places on the tape. After returning later to play the same sections I found they had vanished.
  3. The findings mentioned above sometimes apply to sounds that are originally perceived on the tape. Here is an example with a twist. I was listening to events in real time, as they were being recorded, when a percussive sound attracted my attention. I decided to play it back. It was quite noticeable, sounding like hollow wood being struck. It had a definite pitch or resonance about it, seeming to be one of the strongest I ever recorded and lasting approximately 800ms. I was convinced that anybody I demonstrated it to would hear it as I did. When finally demonstrated it to another listener, he could hear absolutely nothing, even when told later what sound to expect. Also it was noticed that at the exact moment the listener made this remark the event seemed to lose its strength and quality through my pair of headphones.

The above-mentioned effects can only be heard in certain places on the tape every time it is played. It is as if the mental suggestion is finding a section of noise or events that most resembles the frequency content of the hallucination or suggestion.

I devised an experiment to try and shed light on whether the subjective suggestion factor could make one hear sounds that are not there. I employed a microcomputer to store a set of predetermined random numbers which were obtained, not from the microcomputer itself, but from random number tables. The numbers were introduced into a machine code program. The program was then connected to the sound generator in the computer, so that when the program was run random pulses came out of the audio socket. After the program was tailored it sounded very much like the white noise used in EVP, as would be expected. These computer-derived noise events are really pseudo-random, because the predetermined section of numbers repeats itself in a loop format every 15 seconds or so. Ideally it should be as long as possible for this experiment but I was limited by memory space. I recorded about 4 minutes of this on the audio tape. I did not perceive any EVP sounds at all. This means that suggestion or imagination did not build on the random events, whereas in EVP suggestion or imagination seem to build on the white noise. This test, I must stress, is not fully conclusive because 15 seconds of random noise is not long enough. This is because in EVP, using the normal methods, there are dead spots of little or no activity. The implication of this test is that I am interpreting the random events that come from the noise generator, used in EVP work, as containing anomalous non-random events (which theoretically should not be there), caused perhaps by some external force. I either perceive these as they are or by mental suggestion, add to them. When using random numbers from tables, processed by a computer, to produce the equivalent sound, this, in theory, should be exactly the same as the noise generator used in EVP. However, should anomalous events be heard, this suggests that pure suggestion or hallucination is the cause. The test will be made more conclusive in future when I increase the loop time mentioned above to include a larger radom number cross-section. The test so far does suggest that imagination or auditory hallucination does not build on purely random events.

Description of Sounds Heard

In my research there seems to be a certain set of sounds with some individual variation. The strongest sounds often seem to be of short duration, ie. less than 1 second. I would estimate that 95% of the prominent sounds in my recordings are short events. If longer, they seem to be made up of separate events in succession.

There are voicelike sounds which can often be interpreted as male or female, often each with an individual character that will show up on a number of occasions on a section of tape. Sometimes higher pitched voices are heard, similar to those of a child. More abstract sounds are also noted. Common examples are: upward/downward sweeping pitches or tones, short stable frequency tones, all with a certain amount of variation and modulation and finally percussive sounds with pitch content. Returning to the voicelike sounds, the rhythmic structure sometimes resembles human speech patterns, but often a better description would be short, staccato, connected phrases. To me the biggest resemblance to human speech seems to be pitch or frequency content, eg. the tonal sweeps of pitch associated with calling to attract someone’s attention.

The Sounds in Detail

Because the voice sounds seem more complex to interpret, I will firstly be concentrating on sounds that are more basic (as far as measurements are concerned). The simpler sounds, like whistles and tones, should lend themselves more easily to measurement, with a view to future research. It therefore seems common sense to concentrate on the basic events when trying to define subjective effects. It has been noticed that some of these simple tone events take place as a succession of musical intervals, ie. a short sequence of short, stable tones of different frequencies. These sequences are usually one or two notes or events, but up to 6 have been noticed. The larger groups of notes border on simple melody. One can compare the note frequencies using an audio-oscillator (for precise pitch reference) and then come back to the same section of tape sometime later to find the note frequencies the same as first noted. These groups of notes do not change over long periods of time, as far as exact frequency is concerned, even when sections of tape containing notes are randomly selected. This suggests either that subjective suggestion is not significant in this type of event or that people have a good memory for past auditory suggestion. A single beam oscilloscope was used to try and glean more information on frequency components that would correlate with what was being heard on a section of tape containing a six note event. However nothing came to light. I feel that an oscilloscope is not ideal for picking subtle pitch or frequency changes in white noise because of the way it works, ie. displaying amplitude versus time. A more useful system would be amplitude versus frequency - a spectrum analyser. The simple tone events should be more conclusive on the spectrum analyser than a complex sound. This would enable one to put a measurement on the subjective factors.

Testing Human Interaction

I thought I would test a factor which has been mentioned in some EVP texts as affecting results. It is being in the vicinity, trying to affect or even initiate events by concentration on the equipment. I have read that some researchers have found that, if the equipment is left by itself when recording and no attention is paid to it, there are few if any events recorded. I first tried pressing record, vacating the room and occupying myself with some other task. When I returned to play back the tape section I usually found it contained just as many events as normal. After that I thought the next step would be to set up the equipment at a greater distance from all possible observers, ie. a quarter of a mile away in a field. Despite feeling slightly foolish doing this, I considered it necessary as some pattern, related to distance, may surface on analysis. In the experiment I used two tapes. Tape 1 was used in the portable equipment in the field, to record events when activated. Tape 2 was used as a reference tape recording in the normal way, ie. with the EVP equipment in the same area as the experimenter. To try and reduce any suggestion factors ie. one’s predisposed view that one tape will give more events than the other, the tapes’ identifications were masked. After the experiment the tapes were randomised. They also looked identical. I then gave my interpretation of what was on the tapes and after this identified them. The following was the outcome of the experiment:

Tape 1 had 58 events recorded. There were 5 strong events, these being: 2 of the vocal nuance kind, 1 stable tone with a pulse-like quality, 1 percussive sound and lastly a whistle or tone. All the other sounds lacked definition and were very short and fragmentary. The tape duration was 20 minutes.

Tape 2 had 41 events recorded. There were 12 strong events, these being: 9 of the vocal nuance kind and 3 simpler in tone, one being of the twin event tone type mentioned earlier. Again, the other events lacked definition. Tape duration was 20 minutes.

If this test was repeated many times a pattern may emerge. Perhaps these results are indicative of a certain trend, ie. the tape which was in the same room, tape 2, seems to have over 4 times more vocal-type events that tape 1, ie. the one in the field. The test does, however, prove that events can either (and I stress the two different meanings) be perceived on the tape or recorded as events irrespective of the direct involvement of the experimenter. This appears to contradict two researchers’ findings so far.

Of course, with this test and all other tests, it is really necessary to find out more about the subjective or suggestion factors before actual descriptions of the sounds become relevant.

Ideas for Future Research

The need for some visual confirmation of the events heard, ie. the use of a spectrum analyser or similar instrument, is the next step in my research to define the subjectivity boundaries mentioned earlier. The use of an oscilloscope was found largely unsuccessful at giving any detail other than amplitude or level changes on the prominent events. Because multi-person listening gets an agreement that an event has occurred a fair percentage of the time, but fails to come up with an agreement on what was actually heard on a lot of the tape sections, it is difficult to decide whether this is just the subjective factors with an underlying non-random, anomalous event, or both of these plus a third factor - listening technique ie. the training of one’s ears to perceive low level sounds. As I have noticed a low percentage of mutual EVP agreements, this proves the factors are variable. One would assume the third factor to be low, or become low, the more one listens. Research suggests the subjective factors then take over and become the dominant factor in a high percentage of cases. Where there is mutual aqreement the sounds seem very basic. Should it turn out that all the sounds are more simple than perceived underneath the ‘subjectivity blanket’, one could ask, is it possible that without external factors, EVP-type events can occur as unusual but normal phenomena? This is analagous to a succession of lucky dice throws.

In future research it would be nice to measure the subjective factors. One could then test human interaction better if such a factor turns out to exist. If this factor affected the kinds of sounds heard, one could perhaps introduce experiments with a lateral viewpoint.

Author :Tony Bowmaker

Join ASSAP for as little as £5 for Electronic membership

Copyright 2023 © ASSAP is a registered charity, number 327422.