Trayvon Martin, Vigilante Justice

[Natural Born Skeptic]

I have been doing some Googling on the subject of voice identification, in fact it's what initially brought me to this thread. Here are a couple of links so far, for anyone interested in reading them. These are not cherry picked to support my skepticism, in fact I haven't even read them yet.

http://www.sciencedirect.com/science/article/pii/S0167639399000795
http://cdn.intechopen.com/pdfs/16500/InTech-Speaker_recognition.pdf

 

[WildCat]

An example of Tom Owen's views on science and forensics:

Report of the meeting of the SC-03-12 Working Group on Forensic Audio of the SC-03 Subcommittee on the Preservation and Restoration of Audio Recording, held in New York, NY., US., 2009-10-11Guy's obviously a woo pushing flake.

Do you have any examples where he shows he can ID a voice based on a 2 second scream captured by a telephone microphone through a wall at a distance recorded in a lossy format?

No? Didn't think so.

 

[WildCat]

What's wrong with the approach I am taking ?

You didn't accept an extraordinary claim at face value to declare Zimmerman guilty.

 

[Alt+F4]

He doesn't have to rule in Martin. He only has to rule out Zimmerman, since there wasn't anyone else there, and none of the other witnesses claimed to be screaming.

In the recent murder case in which he testified, Mr. Owen made the determination that the the accused was the person who made the 911 call in question based on multiple recordings of the accused person's voice (made two years later). He never testified that he could make this determination based only on a few seconds of a 911 call, something he seems to be doing now.

http://www.stamfordadvocate.com/default/article/Expert-Davalloo-made-911-call-2753115.php

 

[WildCat]

I called for double blind tests pages ago. In any event, the evidence needs to be tested in court.

Because if it's one thing we've learned over the years it's that courts are the best place to determine scientific fact!

That's how we know silicon breast implants cause all kinds of disorders in women, and cell phones cause brain cancer.

 

[WildCat]

According to the web page it has been used in court and helped obtain a conviction:



Just because the supporting research isn't easily found on the Net doesn't mean the testing does not exist.

You forgot to show when he ever was able to match a scream to a voice, particularly a scream captured by a telephone mic at a distance and through a wall.

Will you be doing so?

 

[RandFan]

I have been doing some Googling on the subject of voice identification, in fact it's what initially brought me to this thread. Here are a couple of links so far, for anyone interested in reading them. These are not cherry picked to support my skepticism, in fact I haven't even read them yet.

http://www.sciencedirect.com/science/article/pii/S0167639399000795
http://cdn.intechopen.com/pdfs/16500/InTech-Speaker_recognition.pdf

Thanks. Most of us are incapable of using a search engine.

 

[WildCat]

Thanks. Yeah, I just wanted to see it. I'm leaning toward the idea that this is good science.

Facts:

• The Times Sentinel hired two separate third party experts to do the analysis.
• Both excluded Zimmerman as a match.

For me, here and now, that meets the preponderance of evidence standard in order to form an opinion. But I would love as much evidence as I can get my hands on.

That's the exact result which would be expected if the software didn't work at all under those conditions.

I'm astonished at the complete lack of skepticism here.

 

[WildCat]

Jumping in quite late to this topic: what source of Zimmerman's voice are they comparing the 911 calls to?

The scream heard on one of the 911 recordings.

In every single example this software has been demonstrated to work it compares recordings of the same phrase spoken by people. Not to screams picked up by a telephone mic through a wall at a distance. You have to compare like to like.

 

[Natural Born Skeptic]

Do you have any examples where he shows he can ID a voice based on a 2 second scream captured by a telephone microphone through a wall at a distance recorded in a lossy format?

I can see we are skeptical of this for the same reason. Don't forget, the telephone only captures a very small spectrum of the human voice, and there was other noise happening at the same time.

The fact that this guy claims he can rule out Zimmerman reminds me of TV shows where the crime lab guy zoom zooms in on a license plate that is entirely out of focus and pushes his magic button and all of a sudden it reconstructs the image and the license plate is crystal clear. This "expert" has TV show like powers.

Since I work in the audio field one thing I've learned over the years is that it's much easier for people to understand visual differences than it is audible differences. In this case I don't think people understand just how incredible it is for someone to claim they can identify that the screams are not Zimmerman's. I would be willing to wager a very considerable degree of money that if these experts were put to the (blind) test they would fail miserably.

 

[RandFan]

I can see we are skeptical of this for the same reason. Don't forget, the telephone only captures a very small spectrum of the human voice, and there was other noise happening at the same time.

The fact that this guy claims he can rule out Zimmerman reminds me of TV shows where the crime lab guy zoom zooms in on a license plate that is entirely out of focus and pushes his magic button and all of a sudden it reconstructs the image and the license plate is crystal clear. This "expert" has TV show like powers.

Since I work in the audio field one thing I've learned over the years is that it's much easier for people to understand visual differences than it is audible differences. In this case I don't think people understand just how incredible it is for someone to claim they can identify that the screams are not Zimmerman's. I would be willing to wager a very considerable degree of money that if these experts were put to the (blind) test they would fail miserably.

So, we should simply accept the appeal to your personal authority?

 

[Newtons Bit]

The scream heard on one of the 911 recordings.

In every single example this software has been demonstrated to work it compares recordings of the same phrase spoken by people. Not to screams picked up by a telephone mic through a wall at a distance. You have to compare like to like.

So they're comparing two voices on the 911 call, rather than to a control sample?

 

[Alt+F4]

So they're comparing two voices on the 911 call, rather than to a control sample?

Yes. As Zimmerman has gone into hiding there is no control sample. There were several however in the murder trial in which Mr. Own recently testified. In that case he said it was the accused who made the 911 call in question even though the 911 operator who took the call said:

The voice was kind of choppy and sounded strange," Miller said. "It was a female who spoke, but the sentences were not clear.

http://www.thestamfordtimes.com/story/518494

 

[Moss]

If we take the concept of voice biometrics as valid for a second, a reenactment of that cry for help by Zimmerman would be an interesting thing. Because that should at least improve the odds for a better sample for comparisons.
And I really do wonder why you missed that I am all for scientific testing of that method. I think I wrote that rather bluntly.

 

[WildCat]

The fact that this guy claims he can rule out Zimmerman reminds me of TV shows where the crime lab guy zoom zooms in on a license plate that is entirely out of focus and pushes his magic button and all of a sudden it reconstructs the image and the license plate is crystal clear. This "expert" has TV show like powers.

Yep.

So they're comparing two voices on the 911 call, rather than to a control sample?

He's comparing Zimmerman's 911 call to the scream recorded on another 911 call. I'm unaware of any demonstration where a spoken voice could be matched with a scream, they always require a spoken phrase for comparison, ideally the same exact phrase.

As I said a few posts ago, the result of "does not match Zimmerman's voice" is the exact same result which would be expected if the software didn't work at all under those conditions.

 

[RandFan]

Two independent experts were asked for their opinion. They both excluded Martin. When my doctor, an expert, told me that my problem was anxiety and not heart disease I wanted a second opinion. The second doctor agreed with the first. I could have gone to google and sought reasons to impeach their opinions but I didn't.

I can't dismiss the opinion's of these two experts out of hand. However, I don't put so much weight to their testimony that I would vote Zimmerman guilty if I were on a jury knowing only what I know now.

Now, that's just my reason for thinking it more likely than not that the voice isn't Zimmerman's. I don't mind that people disagree but accusations that I'm not being skeptical are nonsense.

 

[nadsat]

The 2nd expert hand waved the 1st experts' technology though. So what does that mean?

Not all experts rely on biometrics. Ed Primeau, a Michigan-based audio engineer and forensics expert, is not a believer in the technology's use in courtroom settings.

 

[RandFan]

The 2nd expert hand waved the 1st experts' technology though. So what does that mean?

Not all experts rely on biometrics. Ed Primeau, a Michigan-based audio engineer and forensics expert, is not a believer in the technology's use in courtroom settings.

My understanding is that the software isn't the sole determination. In the introduction video the narrator says that further analysis is necessary after the biometric technology. But again, I'm not saying it's beyond reasonable doubt standard. Based on that standard? Not guilty your honor.

 

[Alt+F4]

Two independent experts were asked for their opinion. They both excluded Martin. When my doctor, an expert, told me that my problem was anxiety and not heart disease I wanted a second opinion. The second doctor agreed with the first. I could have gone to google and sought reasons to impeach their opinions but I didn't.

In your example you are assuming that both doctors have experience in determining the difference between anxiety and heart disease. In this case there is zero evidence that Mr. Owen can determine who's voice it is or isn't based on a few seconds of a 911 call. He didn't do it in the Davalloo murder trial and so far there is no evidence that he ever did.

In addition, I'm sure the doctors you would go to would have access to you to examine and your medical history. We have no evidence here that the experts in this case had anything other than a YouTube copy of the audio.

 

[RandFan]

In your example you are assuming that both doctors have experience in determining the difference between anxiety and heart disease. In this case there is zero evidence that Mr. Owen can determine who's voice it is or isn't based on a few seconds of a 911 call. He didn't do it in the Davalloo murder trial and so far there is no evidence that he ever did.

In addition, I'm sure the doctors you would go to would have access to you to examine and your medical history. We have no evidence here that the experts in this case had anything other than a YouTube copy of the audio.

From a "reasonable doubt" standard, given what I know I'm happy to accept your criticism to exclude the testimony.

From a preponderance of evidence standard I would not substitute your ad hoc critique over their expertise. That's an easy one for me. It's simply moving the scales ever so slightly. That could move just as easily to the other side but I would need more than your opinion that the evidence is insufficient. What motivation do two independent experts have to make unsupported claims?