Speech Reversal Lab

REVERSE SPEECH BLIND TESTS

Here's some interesting statistics to ponder. Over the past several months, I've had a blind test on this web site of multiple choice questions. I put this up just to see what would happen. This is of course not official at all and there is no knowledge of the status of those people who voted, nor controls on the tests. But here are the results. Make of them what you will and email me with your comments. They would be greatly appreciated. We are currently in the process of establishing strict testing protocols for Reverse Speech in conjunction with a major Australian University. All comments sent to us would help greatly.

Initial feedback to results of blind test

Subject:
Reverse Speech Blind Testing
Date:
Sat, 13 May 2000 04:27:35 -0700
From:
"George" <******>
To:
"David Oates" <backwards@reversespeech.com>


David:
Interesting questionnaire results. Question 1 can, as I see it, tell most of the story. If there was no difference in the answers people would be guessing and probably getting the "correct" answer about 25% of the time. Given a probability of 25% & a correct response of 284 out of 458 responses gives a very long odds of about ten to the 60th power to 1. That's 1 followed by 60 zeros. If one were to be much more conservative and say that there are really only two answers the correct one and all the others, then the chance of a correct answer could be 50%. So doing the same thing again but with probability to 50% the odds of getting 284 correct out of 458 is only 6,436,000 to 1. This very clearly shows that there is something going on other than guessing. Doing the odds for the whole questionnaire would produce odds that are extremely huge, but not much more meaningful.

What does this tell us? As I see it, it tells us that there is definitely something about the sound of reverse speech that has some consistency. In other words people tend to hear the same thing. That's not particularly useful info though. Analyzing how many people got 10 out of 10, or 9 out of ten could be more interesting i.e. hinting that there is a skill involved.

It seems that you might be wanting to have academic proof for RS. To "prove" reverse speech, It seems
to me that one needs to show that the reverse sounds are meaningful and intelligent. One of the best way to do that as I see it, would be to have RS tell us something that is not presently known, but soon will be known. And particularly something with a probability that is well defined. One of the best ways to get something that is not presently known, and soon to be known, is to have it be a future event. For example: Predict the outcome in advance of the winners of maybe 12 football games. I'm assuming that no one can predict the results better than 60% of the time. Then post these predicted results on the web. Have RS people visit that page and save a copy the predicted results to their hard drive. Making particular note of the date and time so that they can be good witnesses. Then email that they have done so. This will give us a witness list, to prove that these results were done before the games were played. To be considered scientifically valid an experiment must reach at least what is call a "95% confidence level" that means a chance result of less than 1 in 20. At 60% chance of being right, 12 out of 12 would be 99.7% confidence level or about 1 in 460. And 11 out of 12 correct would be a 98% confidence level.

Either should be considered to be a valid positive experiment. With it being a future event, it should be viewed as having proper controls. Given the info that you gave me about a 30 minute tape being required to get a Yes / No or Win / Lose answer, an experiment like this could take analyzing 12 30-minute tapes within 1 week of time or so. I strongly suspect that there are other future events that can be predicted requiring somewhat less time. So this is probably just an example. I believe that this amount of time and effort put into another project would produce tens of thousands of dollars and thus make many things much easier, including proof of RS.
George

Answers as I anticipated they would be.

1/ Get back in it

2/ It broke mummy

3/ They pulled out the grass

4/ None of the above

5/ You were never my Daddy

6/ Hear the lie

7/ Wolf in White van

8/ Who's sorry

9/ None of the above

10/ Cuddle me (but it is an imprecise reversal)

Reverse Speech Blind Testing

Reverse Speech Blind Testing
Current Vote Statistics

Question 1:  Does reversal #1 say? (test1.wav)      votes  percent   1.) Big truck, fix it 18   4%   2.) Get back in it 284   62%   3.) New taxes are wrong 17   4%   4.) No Opinion 139   30%

1 2 3 4 Total Votes:  458

Question 2:  Does reversal #2 say? (test2.wav)      votes  percent   1.) It broke mommy 209   47%   2.) More milk nanny 19   4%   3.) It hurts Annie 37   8%   4.) none of the above 177   40%

1 2 3 4 Total Votes:  442

Question 3:  Does reversal #3 say? (test3.wav)      votes  percent   1.) I saw her for the last time 25   6%   2.) They pulled out the grass 145   33%   3.) Hey, you took all my brass 20   5%   4.) None of the above 247   57%

1 2 3 4 Total Votes:  437

Question 4:  Does reversal #4 say? (test4.wav)      votes  percent   1.) I love the parks. We took a friend 18   4%   2.) We hunt for sharks. Who stole my pen 13   3%   3.) Lover parts when we shot at them 277   63%   4.) None of the above 131   30%

1 2 3 4 Total Votes:  439

Question 5:  Does reversal #5 say? (test6.wav)      votes  percent   1.) We cared more for the fatty 14   3%   2.) I even hated my mommy 25   6%   3.) You were never my Daddy 263   60%   4.) None of the above 134   31%

1 2 3 4 Total Votes:  436

Question 6:  Does reversal #6 say? (test7.wav)      votes  percent   1.) Fear the lie 50   12%   2.) Hear the lie 152   36%   3.) Hear the life 66   16%   4.) None of the above 154   36%

1 2 3 4 Total Votes:  422

Question 7:  Does reversal #7 say? (test9.wav)      votes  percent   1.) Wolf in white van 187   44%   2.) Vets still in 'nam 27   6%   3.) Hex on white man 38   9%   4.) None of the above 169   40%

1 2 3 4 Total Votes:  421

Question 8:  Does reversal #8 say? (test14.wav)      votes  percent   1.) Who's sorry 195   46%   2.) Noose funny 27   6%   3.) Why worry 23   5%   4.) None of the above 180   42%

1 2 3 4 Total Votes:  425

Question 9:  Does reversal #9 say? (test15.wav)      votes  percent   1.) One can in the office was there 24   6%   2.) The crime of love now worsens 148   35%   3.) Samson came on a very thin thread 10   2%   4.) None of the above 241   57%

1 2 3 4 Total Votes:  423

Question 10:  Does reversal #10 say? (test19.wav)      votes  percent   1.) Dad will see 33   8%   2.) Cuddle me 122   28%   3.) I hate food 14   3%   4.) None of the above 264   61%

1 2 3 4 Total Votes:  433