|
|
|
|
THE CASE...
In the meeting of ANT - National Association of Transcomunicadores happened in 6 September/98, we counted with the participation of 97 experimentators of several regions of Brazil, that witnessed the phenomenon of contacts with the Spirits in hundreds of occurrences.
66 questions were put by these friends to their deceased loved ones - and we received 400 answers approximately, between direct answers (by those deceased ones) and comments, from the part of the Spirits Communicants.
Obs.: Part of this results are listed in our recent publication (Newsletter 39) offered in this site.
As we are already accustomed, the quality of those contacts are usually "A" and "B" type, that is, with perfect audibility.
Due to the clarity of many contacts it was necessary to introduce the practice of registering the recording with several sources, in other words, in that meeting we recorded with 5 tape recorders and computers, to make possible the subsequent comparison among the contacts received.
Obs.: due the fact of many voices being so clear (class "A" type) it is not easy to indetify what are HUMAN voices and what are SPIRIT voices. Only having more than 2 recordings that we can compare, because the HUMAN voice will be always the same in all recordings though, Spirit voices will only appear in one isolated recording.
It is this way that, a lot of times, we noticed that certain voices are, in fact, paranormal, because they appear in a recording and not in the other, made simultaneously.
Those explanations are necessary for us to approach the case that we will analyse here.
AS IT HAPPENED...
In the opening of the meeting, I listed some friends' names, there presents, who cooperated in the organisation of the event.
One of those mentioned people was Romeu Mandato, that always aids in the transport of those friends who come from outside of São Paulo for our meetings.
When I mention and I thank him, a paranormal, class "A" feminine voice complemented the speech like this:
Sonia: - " Thanks to Romeu for the help... [feminine paranormal Voice: - " AND FOR the DEDICATION "] I Also want to thank to the ones who came from far to our meeting..."
It happened that just in that beginning of the meeting, we had only two tape recorders registering simultaneously and we noticed that just in one of them appeared that voice, that sounds, in fact, different from mine.
Nevertheless, we considered this not enough as a proof, since the voice is as clear as mine. So, I doubted myself: did I say that? Was it a spirit voice?
We had to investigate.
We requested a specialist in analysis of Voices from USP - University of São Paulo so that he compared all the recordings and investigated if there were two different speakers in fact (A and B) or if just one (A = me).
That analysis would also implicate in the following:
if there were two speakers, one of them would be paranormal - because only one person was in front of the 97 presents. So... if there existed ONE person and TWO different voices... the second one had to be paranormal.
Listening to the whole sentence we perceive that both voices sound
a bit different, but we had to prove it; only a experienced analyst and
real expert in voices analysis could attest yes or not.
 |
Here is the Technical Declaration concerned the above audio sample, which follows the international pattern of analysis of voices which can be repeated by any scientist.
WHAT WAS SUPPLIED TO THE EXPERT:
As we had to check if the voice was mine or not, I sent to the analyst
. the original recording
. plus one I made with my natural voice repeating the whole sentence.
This way he would have elements to compare details of my voices and the "unknown" voice.
The expert called these recordings as TAPE 1 and TAPE 2 (being the "1" the original and the "2" only my real voice).
Pieces of Exam:
There constitute pieces reason of the present exam two recordings, numbered 1 and 2, recorded in the same equipment, digitalized in a standard computer IBM-PC through a standard sound card Sound Blaster. The used resolution was of 16 bits, with rate of sampling of 11.025 samples per second.
Objective of the Exam:
To accomplish the verification that both recordings were made by the same speaker (or not) from now on denominated speaker "A".
Patterns of Confrontation:
There are object of patterns of confrontation 4 numbered recordings in the following way:
Previous processing of the Samples
The samples were normalized in way they could have the same sign level and the same relationship sign-noise, the spaces without voice were removed, since they don't characterize the speaker. Of the normalized samples there were extracted coefficients of the type MFCC (Mel-Frequency Cepstral Coefficients) using windows of 23,2 ms with spacing among windows of 10ms, 5 filters for octave, resulting in 13 coefficients per window.
Methods of Comparison:
The coefficients of the sample "tr1" were used for training of a system of automatic recognition of the independent speaker of the text based on Gaussian mixtures (GMM - Gaussian Mixture Models), using 24 mixtures to represent the speaker.
As algorithm of comparison it was used the logarithm of the verisimilitude (log-likelihood). This method was chosen because of its consistence, not being the only one possible to be used.
Results:
The obtained results are compiled below in the table:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The number of vectors indicates the number of resulting windows of the pre—processing; the output value it is the sum of the logarithm of the probability of the vector to belong to the speaker in subject; and the normalization is made dividing the output value by the number of vectors.
Conclusion:
Through the obtained results we conclude that it doesn't exist evidence that all the locutions/speeches have been pronounced by the same speaker ("A"), because the normalized values are not close enough.
OUR COMMENTS:
We know that Technical Declarations are of difficult understanding, so we summarize here that, through the elaborated calculations, the specialist concluded that... yes, they are two different speakers in the recording.
FINAL CONCLUSION:
That Declaration plus the fact that we have two simultaneous recordings in which only in one there appear a voice that says [-"AND FOR THE DEDICATION"], it is a contusing proof that, in spite of clear and high, the voice in fact belongs to a spiritual communicant.
For whom imagines that the contacts of TCI are always
noisy and of difficult perception, here is a contrary example.