Telephone Conversation

between Naveen and Jack Mitchell (Forensic Examiner)


Naveen: Okay. And we are doing this recording by mutual agreement. Jack: Right

: Anyway thank you for giving your time and I understand that you are doing this on a Saturday and I appreciate your giving us the time to do this. Jack: Well, that’s perfectly fine.

: So now regarding the work that you’ve done, could you please describe what kind of work you do as a professional in this field.

Jack: The broad category would be audio engineering. Specifically what I do is forensic examination. I specialize in that, although I do do some other work. I will do restoration of old recordings…old 78’s say out of the 30’s and 40’s. But 95% of my jobs are forensic examination.

Naveen: I see. And how long have you been doing this kind or work? Jack: Since 1995.

Naveen: Okay. Could you describe or mention some of your other clients…your past clients?
: Sure…various agencies of the United States Department of Justice and that includes the Drug Enforcement Administration, United States Attorney’s Office, the Civil Rights Division. There have been other law enforcement agencies…municipalities law enforcement agencies, public defender’s office, various attorneys from all over the country, insurance companies, and private individuals who just need some answers for one reason or another.

Naveen: Just ballpark…how many clients would you estimate that you’ve had in the last five years?
: (laughs) I haven’t the foggiest idea. (Naveen: Hundreds?) Hundreds? (Naveen: or dozens?) I don’t know that I would say it was hundreds, but it’s certainly enough that it keeps me busy. How many of these jobs can you do in a year? Some of them are quite involved. I just recently did a job for the Department of Justice that took three weeks…well there are just 52 weeks in a year, so you can’t fit that many clients in.

Naveen: Okay I get the picture. How do your clients typically use your work?
: Well they’ll use it for a number of different things. A lot of the work is enhancement…that they need to enhance the recording so that it is intelligible so that they can get official court transcripts off of it. And we do that in a number of different ways. I will work with the tape for whatever it takes till I get it to the place where I think it’s good enough. And then sometimes they will actually send in a certified court reporter right into the studio environment so that they can get a very accurate transcript. That’s done because I can do things within the studio that I can’t give to them on tape or CD. These are things of putting things on loops where we take a small segment of the audio of just a couple of seconds or a couple of words. We put it on a loop and it will loop around indefinitely until we tell it to stop, and while it is looping we will make certain adjustments to it that will adjust to the hearing ability of the court transcriber. We also do things slowing it down and using pitch correction and we also will take a process file and the unprocessed file and feed the two together. These are things that I can’t do when these things are sent back to the client. So that is one way that it is used.

Jack: Another way that it is used it to document certain events on a recording. Those events may be stop/start routines of a recorder where the opposition may be claiming that the recording is falsified in some way. And depending on which side I’m on, I may be hired to either prove that or disprove it. And the way we do that is looking at the waveform and through spectrographic analysis of the signal as well as doing magnetic tape development and looking at the tape under high magnification. Most of the time that sort of thing is fairly easy to detect because there are gaps in the tape and if somebody will claim that the tape is a continuous tape and there are these start/stop routines, it can’t be a continuous tape. And we’ll also look at changes in frequency on either side of these things to try to determine if it was some sort of recorder anomaly or if it was truly a stop/restart.

Naveen: What about the nature of the work you did where you were actually trying to establish what was being said in a whisper…how would you describe that kind of work? Jack: Difficult. What do you mean by how would I...

Naveen: What category would you put that in? You were describing the kinds of work you’ve done for your clients... Jack: We call that speech decoding.

Naveen: Speech decoding…okay. And how does that…what is the science or the technology behind that speech decoding that you do? Jack: Well we use spectrographic analysis for that. Spectrographic analysis…let me explain how this works as best I can. A spectrogram does not tell us what is said. The conclusions are based on the combining of an aural sense, a visual sense, and the technical data that us presented by the spectrogram.

Naveen: Just like a multiple side analysis then?
: It’s a multiple side, yes. But it’s all combined into one. It all happens simultaneously. What we do is we have the audio file and we always use an unprocessed audio file. We do not process that file in anyway because we could add or subtract important data. And this is then put in to a spectrographic program that will show a whole lot of things. We have what we call a FTA display which is frequency, time, amplitude…whereas the frequency is on a vertical scale and the time is on a horizontal scale, and the amplitude is shown in both the vertical and horizontal, but it is shown as brightness or dullness of color. The brighter the color the more amplitude that we have. We can also do what we call LPC form and tracking in which we can actually track the movement of the tongue. We use amplitude envelopes to show us how many syllables are involved. And FFT fast (?) transform analysis which allows us to look at very small segments of the signal to see what sort of frequency content is there. Is the signal a noisy signal such as a fricative or is it not as noisy such as a vowel?

And we put all of this together…we have it in front of us…we have the waveform lined up exactly with the data that is being displayed. And we are able to segment the information into really very small components. We can segment it down into segments less than one nano (?) second which is actually not useful at all because everything disappears at that point. But we can segment it into useful segments of milliseconds. One millisecond might be useful or three milliseconds might be useful. And just look at that. And we’re able to align…we visually and orally (or aurally)? can align what is happening in that waveform with the data up into the display of say the frequency time amplitude display. That shows us where things are happening and can confirm or disconfirm what we believe we are hearing. We can look at, depending on what settings we use, how big of a display we make…we can look at a frequency time amplitude, an LPC, an amplitude envelope, an FFT and the waveform all at the same time. So all of these things line up vertically and horizontally.

Naveen: Very interesting. Could you very briefly describe the technology (hardware, software ...maybe the dollar value of the system that you have used in the work that you did for our particular case)? Jack: Well... as into particular brand names, you mean? Or…

Naveen: It doesn’t have to be brand names, but generally…
: It’s digital’s computer-based…and there is a very high quality audio card inside the computer which is what we call the IO... it takes it in and puts it out. There’s a high quality interface that interfaces with that…that will take an analog signal and convert it to digital, and vice-versa... digital back to analog.

Naveen: What about the dollar value…what would you say the dollar value for everything that you put together to do your kind of work that you have been describing?
: My system is probably ...I can’t give you exactly, but I think it is somewhere between $50,000 and $55,000 dollars. That sort of constantly changes, as does the software. The software that is used, that I use, is very high quality analysis software. Actually for the spectrogram what I use is two different analysis programs. One of them is called Soundscope which is a very highly regarded program that is used in universities and hospitals and private research facilities all over the world. The other one that I use is a program called Signalize which had gained great favor with the academic community. The work that I did on this project that we’re talking about was put into both programs. The primary program would have been the Soundscope and then I would put it over into Signalize to see if I got exactly the same thing, which I should…one would confirm the other.

Naveen: Would you say that your equipment then ranks with the equipment that’s used by top audio(?) in the world? Jack: Absolutely…absolutely. Now what might be considered state of the art tends to change from minute to minute in this day and age. The fact of the matter is that working with these kinds of signals does not really require anything beyond a reasonable audio board in a computer. The thing that you need to be concerned about is the input and output signal to noise ratio. And it is generally considered that a signal to noise ratio of 87 dv is quite good. My audio board has a signal to noise ratio of 105, or maybe it’s a 104. The bigger the number the better. What that means is that any noise within the system is down so low that it is not perceived. That’s one of the primary considerations. There are an awful lot of programs that can do the filtering and compression that we use. Some programs are more precise than others. But the thing that is really important is the knowledge and skill of the investigator…important beyond a certain level of equipment. I have been told by others that my system is way beyond what is needed for this kind of work.

Naveen: Okay. Well moving on to the discussion about the work that you have done. How did you receive these tapes and what instructions were you given? Jack: Well, by how received do you mean who?

Naveen: Who sent them to you and what were you asked to do? What were you told about...
: I was contacted by Randy Stein. I think it was early December 1997…it may have been late November or early December and I talked with him and then there was no commitment from him to hire me. I got a call somewhat later, I don’t know if it was three days or a week and a half, I don’t recall at this point, but he wanted to retain me to work on this. What the definition of the word "this" was, I didn’t know. I actually asked what this was all about, and I was told that because of the nature of this (other than giving me sort of general instructions as to they wanted enhancement of certain areas of the tape recording), other than that, I had no idea what any of this was about. They felt that this was important. Frankly from my end I didn’t think that it was because as soon as somebody tells me that they think that this or that says such and such, I turn around and try to disprove them. But he felt that he didn’t want to tell me so I…you know, that’s fine. But he sent me the first tape. As I recall the first tape was very degraded. It really was not a good tape to work from and I told him this and then he got another tape that was of higher quality. And it was a tape that supposedly was a first generation copy of the original. It was not the original but a first generation copy whereas the other one sounded to me like it was a fourth or fifth generation copy---it was so bad. It was much better than the one I previously had, although it was not the original and was still somewhat degraded.

Naveen: Okay, so you were not told at all what was the nature of the investigation or...
: No, they wanted me…they told me certain timings on the tape…to pull those out, try to enhance it, and let them know what I thought was being said.
: Okay, then the procedure that you used to work on this is what you described a few minutes ago: the multiple-sided analysis...
: Yes, the spectrographic analysis.

Naveen: Basically you were saying four or five things: the amplitude envelope, the spectrographic, the starting and stopping points were talking about four or five different manners or ways in which you look at these segments? Jack: Right, the FTA display, which is what people generally call a spectrogram. A spectrogram can actually be any sort of display of the spectrum. It can be an FFT display, or even the amplitude envelope. But it would be the FTA display which shows us the formats (?) of a person’s voice. And the stop/start points and all of that would line up with the waveform as well as our oral perception…we would segment. We could segment into a partial word, into one syllable, into one word or a partial phrase or listen to the whole phrase ...figure out where things would start and stop. Naveen: Well if you like, you could discuss this further otherwise you had only talked about this a little while ago so I won’t ask you to repeat that. But feel free to say more if you like. Now you did send a copy of your report recently to Bill Ogle…do you stand firmly behind your conclusions in that report? Jack: Absolutely. There has been nothing presented to me that would change my mind even slightly.

Naveen: So I’m just going to take that your conclusions are documented. I won’t ask you what those were because there is no need to talk about that. Jack: Right. My conclusions are documented in the reports that I have presented.

Naveen: Did anybody else review your work and agree with you or disagree or collaborate with you in any way, any other person with professional qualifications? Jack: Well with the spectrographic analysis I felt that it would be best…that it would be best to bring in a consultant who had a good strong background in speech science. And so I consulted with a Dr. Helen McCafrey(?) in the Department of Communication Sciences and Disorders at Texas Christian University. I sent her the audio files of the various segments. She has the same software, the Soundscope software, and after she got the files and had the opportunity to set up and I think she reviewed them first. Then we had a nine hour telephone conversation in which we analyzed every one of the segments that was requested. And we would go back and forth as to the various segments, segmenting it, looking at the various components of it. And working with each segment until the two of us were in agreement as to what we thought was said, what was the most likely that would fit in to the particular formations that we were seeing, the data that was being presented by the spectrogram.

Naveen: And what were her conclusions after the nine hours of working with you? Jack: Well she was tired, I can tell you that. But her conclusions are the same as mine. I then wrote the reports, documented everything. I sent her the report for her review and told her that if there was anything in there that she felt was incorrect she should tell me she should change it. Or if she felt that anything needs to be added, then she should add it. She added nothing; she changed nothing. She actually submitted a document in which she says that she is in total agreement with the report with the findings.

Naveen: Does Bill have a copy of that document also, if not...? Jack: I don’t know.

Naveen: Okay, I’ll ask him if he’s got one otherwise we may... Jack: I don’t know. I’ll tell you it is… This book, "Someone Has Poisoned Me" was sent to me by Nico. It is in there. It’s on...

Naveen: By the way since you brought that up, when did you get that book and learn all the other things that that book kind of talked... Jack: Right after it was published last spring.

Naveen: Okay, so it was way after you had finished your work. Jack: Yes. That was page 310…is what she submitted. Let me state right now that it is exactly what she submitted. It is not altered in any way. And while we are on the subject I will make that same statement for the reports that are published in here that I submitted to Nico and Randy Stein.

Naveen: So I’m just assuming, I haven’t asked them, I’m assuming those two then decided to collaborate at a certain point in time as far as your work went at least ...Nico and Randy? Jack: Well as it turns out, I believe, they were collaborating from the beginning…that Randy was what we would call the front man and Nico was behind the scenes. Eventually Nico just took over in having direct dealings with me. (End of side A of tape)

Naveen: You were saying? Jack: Have you turned the tape? (Naveen: Yes) We were speaking about the book "Someone Has Poisoned Me" and I was saying that the reports that are published in this book are exactly as I submitted them. They are not altered in any way.

Naveen: Okay, that’s as far as the publication of that book goes, then. Jack: Now I will say that the full report of the two ...there were two parts to the study: an authenticity and the analysis of segments. Not everything, not all of the segments have been presented because they turned out to not be relevant in that people thought that certain things were being said but the conclusion is that they were incorrect and really had no relevance to the investigation. The full report is, as I recall, was roughly 100 pages long of which I believe 50 pages that is relevant to what this book presents.

Naveen: I should ask you, is there anything in the report that would go against your findings or conclusions or weaken your conclusions in any way? Jack: Anything in the ...what?

Naveen: Anything in the rest of the report that was not published? Jack: Oh no, no. Absolutely not. No, I mean what wasn’t published was that there were a number of segments that were defined for investigation. And what was being said in these other segments were not relevant…they had no relevance to the problem at hand, which was the word "poison". After I got to a certain point in the analysis I was told what was going on. I had already on my own come up with the two segments in which we found the word, some form of the word, "poison". I believe I had done the other segments also in which they seem somewhat innocuous, although I guess that depends on how you view that, but I know that some people thought that some of these other segments had the word "poison" in it, but they do not. And I think that...

Naveen: Okay, so later on they told you they thought that the word "poison" was other places and you disproved that or you didn’t agree with that? Jack: Well my conclusion is contrary to what they thought was there. I think there were five segments and two out of the five came up positive, and the other three are negative. Put it that way.

Naveen: Okay, that’s a good summary. Were you contacted by anybody else regarding your work other than Randy, and Nico and myself and Bill Ogle? Jack: Well, I’ve had some calls from Australia. I had a call from an individual sometime last fall. I cannot remember his name. All he wanted was to get the original audio files and I told him that I couldn’t do that. You know, he wasn’t my client, but there are these files that I posted on my Internet site, my web-site that are MP3 files that I would be happy to send him those, which I did. And that was essentially the end of that. There wasn’t any real discussion about anything. In the last couple of weeks I’ve had phone calls from a Mr. Hooper who has had some questions and has expressed some opinions. He is having difficulty with what we call segment 1, phrase 2 in which we say the dialogue is "the poisons (?) (poison’s)? going down." His difficulty with that is that he does not perceive the word poisons. He seems to agree with the word "the" and he seems to agree with the words "going down", but "poisons" he cannot agree. Now the first time he called me he said that what he hears is "boys are".

Well, the formations that we see in spectrograms can be caused by other letter combinations. What we have focused on in that was the "oi" diphthong. Now "oi" is oy. Well that is the same sound that we get when we say the word "boy". So up to that point could it by "boy"? Well, perhaps but now how do you account for the double sylballents that we can document? Well, he said, "The boys are going down." That doesn’t fit. It doesn’t work in the spectrogram. Could it be, "The boys is going down"? I don’t think so. First of all that’s bad English. That would account for two sylballences, but in the section where we get "on" for poison there is an energy drop: if we have "boys is" we would have no energy drop in the lower frequencies. So I don’t think that that is particularly a credible alternative to this.

He called me again a week ago. He said that he had gotten a recording from the archives, as a first generation copy, and now he is absolutely certain that what is said is "The swelling’s going down." Well, I don’t believe that is a credible alternative either because the "w" "e" will not give us in the spectrogram the type of formation that we see which is a movement in the frequencies from 800 up to about 1600 from a left right direction. The "w" "e" in "well" doesn’t work. Now if it were "we" just the word we then I could say that could be credible under normal circumstances we would see a faster rise from 800 up to the 1600. But it could be "we", but how does that fit in: well, it doesn’t. It could be "wheel" well that doesn’t fit into context. It could be "weasel" or "weasels" that would take care of all of it, but what did they say, "The weasel is going down?" That doesn’t make any sense.

Naveen: Did you do by the way, any kind of dictionary search to see if possibly the word "poison" was something else when you were considering all different possibilities?

Jack: Well, we went through the process that you and I are going through right now as to what could be alternatives. You know, I had already thought of "boy" but that doesn’t account for the rest of the contents of the spectrogram. You know, it could be "ploy": well, I don’t think so. That doesn’t account for the rest of the spectrogram. I actually sat down with this Random House unabridged dictionary that I have and tried to think up every alternative that I could in every letter and started leafing through the dictionary trying to come up with any kind of reasonable alternative. And I could not do it. What this keeps doing in everything that I could think of and every alternative that has been presented to me is that it keeps going in full circle, and still comes back to the only thing that fits in there with the spectrographic display and the oral sense is the word "poisons".

Naveen: So you have not been able to come up with any other word except the word "poison"? Jack: That’s all I ‘ve been able to come up with. Does boysenberry fit in there? That doesn’t make any sense. So I think that our conclusions…well I have this book sitting here, let me look at the report and see exactly how this was worded…if I can find it. On page 300 right at the bottom of the page it says, "Note, all features isolated and evaluated are consistent with "poisons". Analysis data combined with oral perception indicate it is more likely "poisons" than any other English word." I still stand by that today. Now if somebody comes up with a viable alternative that can fit into this spectrogram: it can fit into the FTA display, it can fit into the amplitude envelope that the LPC form of tracking is consistent…then I’m going to be the first one to say so. But so far nobody has come up with anything.

Naveen: Let me ask you, are you prepared to, if necessary, do a video tape report about this work if necessary in the future? Jack: I think that I would be happy to do so. It’s something that we would have to discuss the format of it and so on. But I would have no objection to that. I stand behind this report 100%. Now there’s always the question of what can be challenged in a report. And we have no idea what an opposing side may choose to focus on and challenge. So there may be something in the report that they want to focus on and challenge and I don’t know what that might be. But I do stand behind it and… The only thing that I could think of that people might want to challenge is this segment 1 because of the audio being so degraded.

People tend to want to believe their ears only. Since this is degraded audio it is whisper production and as such it is defined as being distorted vocal production. It creates auditory illusions, and when a person listens to this kind of audio at one point they think it says this or that. And they are very sincere in what they believe. And then they listen to it an another time and suddenly it sounds like something different to them. And so I suspect that that is going to be a point that would be challenged. And I suspect that this particular segment may be controversial forever. However, there are other points of evidence in this report so…There are also variables when people listen to this audio. Not all audio playback systems are created equal I’m afraid. And some of them will add their own coloration, their own artifacts, speakers are notorious for doing that. Also, you have room conditions in which the listener is in which can add reflections. Not too long ago, the last…I guess it was October…I did a restoration of an old 78 and the client listened to it on one system, and said, "Well where…you know, can’t you get that particular anomaly out?" And I said, "Well, it is out. Come and listen to it on this system." So we listened to it on a different system and it wasn’t there.

Naveen: I was going to ask you next, would a session in your studio help with the aural hearing of these segments because of the quality of your equipment as compared to somebody’s home stereo system or something like that?Jack: Well this is why attorneys really like sending certified court reporters into my studio because the quality is so much higher. The studio is acoustically designed by a professional. It has all kinds of absorbers and reflectors in the room. This segment is still somewhat difficult to hear, but it’s a better environment than most. However, I have played this, since all of this was released, I have played it for other people and outside of the studio environment. As a test I made a CD of it. Took it over to a friend’s house and put it on their $500 system and my friend’s wife who didn’t know anything about any of this, picked it out right away.

Naveen: Picked out the wording as you describe it? Jack: Yes. She had to listen to it two or three times then she picked out the wording. So there are going to be individuals who are never going to understand the relevance and importance of the spectrogram and they’re going to believe only their ears.

Naveen: So when you said earlier that the controversy may go on, that is only based on audio and hearing, not the controversy of your report itself? Jack: Well I think that the report is extremely solid. I think it’s going to be extremely difficult to discredit any area of the report. I think that if there is one area where they will try to discredit, they’re going to try to say that what I hear is such and such, and that’s different than what you’ve reported. So your report can’t be correct.

Naveen: So that’s why you were describing how the hearing perception changes from day to day. Jack: Right. We call this auditory illusion.

Naveen: So that’s why doing all this other kind of work that you’ve done is so critical to coming up with the correct analysis. Jack: That’s right. That’s right And it is my opinion that the spectrogram locks this in. And with trying to come up with alternatives, and not being able to, and nobody else coming up with an alternative that is viable. I think it is pretty well locked in to…as presented in this report that it is more likely the word "poisons" than any other English word, and I stand behind that.

Naveen: Okay, Jack. Well, and you know we’re not talking about the other segment which I’m assuming because it’s documented in and reported, you stand behind the other segment also. Jack: What we call segment 4?

Naveen: Right Jack: Yes, absolutely. So far I have not even heard any controversy about that. Nobody has called me and said, "Well, I hear something different there." That’s really much clearer. The voice production, although it was somewhat distant from the microphone, it was normal speaking voice.

Naveen: It was not a whisper. Jack: It was not a whisper. It is really quite clear and I think anybody would have a very difficult time challenging that.

Naveen: Well I really thank you for your time and I’m hoping this recording came out okay. Let me hang up and I will check it real quickly and make sure it’s okay and if so then that conclude our discussion for the day. And we’ll see where this goes from here. Jack: All right, well you never know. I mean, the opposition certainly does have the right to oppose and present their case.

Naveen: I’ll just go ahead and stop the recording if that’s okay with you and we can just say our good-byes for today. Hold on please.