A Basic Experiment in the Electronic Voice Phenomenon
Introduction: In the this experiment a short session was carried out using the prototype Alpha apparatus prepared for the TV visit last summer.
The system uses a radio which picks up the large amplitude signal from the Alpha and a microphone which picks up the acoustic output from the radio's loudspeaker.
Prior Assumptions: It was said by Meek et al, (and seemed to be proven in our own experiments), that an e.m. "gap" and an acoustic "gap" aided in getting a comprehensible signal.
It was assumed that the output of the Alpha was providing acoustic "raw-material" partly formatted ("transformed") into the required form, the formatting being completed in the acoustic phase.
It had also been shown - by the use of acoustic attenuation means surrounding the microphone - that the EVP signal was acoustic in nature since its amplitude was inversely proportional in some way to the amount of sound insulation used to isolate the microphone.
Preparation: In this experiment the voltage across the loudspeaker voice-coil was monitored and recorded on channel B of a stereo recording.
The output voltage from the microphone was recorded on the stereo "A" channel.
Prior to the experiment the radio was tuned to a broadcast station, and the independently adjustable gain controls on the stereo channels set to give equal indications on the bar-graph displays.
Results: a 2 minute session was carried out with (relatively) poor results (as is normal when starting up a rig (hereinafter "EVP Station") for the first time.
As mentioned above, the stereo recording consisted of a track related to the voltage across the radio's loudspeaker, and a track related to the electrical output of the microphone. The B track is the stimulus and the A track is the response.
Analysis: One of the utterances heard on the tape, 'Morris' , was selected out for further study. Figure 1.
In the following Figures the lower trace is the voltage across the loudspeaker, which then results in the upper trace, which is the voltage produced by the microphone when "listening" to the loudspeaker. The lower trace is the cause, the upper trace is the result.
Then the vowel section where the 'o' changes into the 'r' was selected out and is shown in Fig. 2 above..
As can be seen, particularly in the first three sections, the idea of a roughly formatted wave being finally transformed into a complex and comprehensible form seems to be borne out. You can see the rough bumps in the lower channel seem to be transformed into the much more meaningful upper waveforms which correspond to a large extent to the usual form for a vowel sound.
It is rather like vocal cavities being stimulated into resonance by pseudo-glottal impulses.
The first thing to be considered was room resonance and so an additional test was carried out wherein acoustic white noise was injected into the room and the results monitored and recorded via the same system as before. It was found that there were no significant spectral peaks resulting from room resonances. We can forget about the room adding anything to the upper trace.
The whole analysis was turned upside down however when both channels in the above instance were subjected to band-pass filtering limited to the basic speech band, see Figure 3 below. (The bottom signal has been multiplied by +6dB to enable easier comparison.
Looking at the spectral plot - the spectrum of the voltage applied to the loudspeaker is in purple, the spectrum of the voltage produced by the microphone is in light blue.
As you can see, the microphone voltage tracks the spectral response of the voltage applied to the loudspeaker very well, except for what is probably a slight acoustic resonance (probably due to the speaker and/or room) just below 1 kHz.
Compare with the spectral plot before filtering, the spectral plot derived from Figure 2.
Using a publicly available Speech Synthesizer the PC was made to speak the word 'Morris'., and then the 'o' portion of Morris was displayed as shown below, Figure 6.
Conclusion: That in the Alpha Interface System the acoustic elements play an essential but passive role in the production of EVP - the essential part being due to producing an optimised spectral response from the viewpoint of audibility and intelligibility, the direct electrical signal being prone to masking by the larger low frequency components.
Caution: It should not be forgotten however that an operational EVP Station will also be prone to record pure audio EVP. This will be the next subject to be addressed.