Trayvon Martin, Vigilante Justice

[RandFan]

12.35% + 48% = 60.35%. So there you go people. Tom Owens $5000 piece of software confirms there's a possibility it is Zimmerman's voice.

Uh no. "False rejection rate" does not mean "statistical margin of error" in the sense you are trying to use it. FWIW: of course there is a possibility. 48% = possibility. Did you mean to say that 60.35% means probability as opposed to possibility or did I fall for a joke?

 

[Skeptic Ginger]

Hmm . . . I think if this were an issue brought up in court Owen's status as an "expert" would be considered valid but he would be completely destroyed by a defense expert. Owen's actions are actually contrary to his own earlier writings and common sense audio forensics practices. ...

He's not a politician being accused of flip flopping. Scientific fields are known to advance and juries are capable of understanding that.

 

[Professor Yaffle]

Uh no. "False rejection rate" does not mean statistical margin of error. FWIW: of course there is a possibility. 48% = possibility. Or did I fall for a joke?

I'm not au fait with voice analysis, but I'm not sure that the 48% refers to the liklihood that Zimmerman is the voice on the tape. 48% is likely to refer to the percentage of parameters that matched which is then converted to a likelihood of a match and is probably not linear as there will be some parameters that are shared by lots of people and others that are more distinctive. I don't think nadsat's adding up of the percentages makes much sense either.

FWIW I think he would be doing his credibility harm if he is overstepping his expertise here, and this would ultimately harm his company more than the publicity would help him.

 

[RandFan]

I'm not au fait with voice analysis, but I'm not sure that the 48% refers to the liklihood that Zimmerman is the voice on the tape. 48% is likely to refer to the percentage of parameters that matched which is then converted to a likelihood of a match and is probably not linear as there will be some parameters that are shared by lots of people and others that are more distinctive. I don't think nadsat's adding up of the percentages makes much sense either.

Mea culpa. Yes. However, once converted it likely would lead to a degree of possibility even if it s not a 48% possibility.

 

[Professor Yaffle]

Mea culpa. Yes. However, once converted it likely would lead to a degree of possibility even if it s not a 48% possibility.

Presumable a much lower one than 48%. I doubt he would be excluding Zimmerman if the probability that he was falsely excluded were anywhere near that sort of figure.

 

[Kevin_Lowe]

Here is the CNN video.

http://videos.mediaite.com/video/Audio-Forensics-Expert-Tells-CN

Curious when he compare the exact same clip to each other it only come back with a 99% match. Also, note the "false rejection rate" of 12.35%.

12.35% + 48% = 60.35%. So there you go people. Tom Owens $5000 piece of software confirms there's a possibility it is Zimmerman's voice.

It looks like reporters are mixing up "48% match" and "48% chance it's the same guy".

If you take 100 measurements of different parts of my body, and the same 100 measurements of Keanu Reeves' body, and 48 of the measurements match (to within the margin of error of the apparatus) that doesn't mean there's a 48% chance I'm Keanu Reeves. Although on the internet you never know.

From the additional context provided it looks like they actually take a 48% match to be significant evidence against the hypothesis that the voices belong to the same guy.

 

[Natural Born Skeptic]

And let's not forget that self-taught (ring a bell?) fire/arson "expert" Manuel Vasquez got Cameron Todd Willingham sent to the death chamber on "science" since revealed as complete garbage.

The thing that is utterly hilarious here is the people insistent on defending Owen either have no clue how easy it is to become an "expert" or don't care one way or the other.

Here's how to become an expert of this type:

1. Take a course for a few days and you're a "voice identification expert".
2. Now if you know what you are doing, you get heavily involved in all of the industry organizations. This gives you cred and you start to get called as an expert witness.
3. Now write papers for organizations that have much, much lower standards that peer reviewed scientific journals, and are easy to get published in.
4. Now start to promote yourself in the media.
5. Do it for many years.

And bingo, you'll have people on JREF swearing your a established expert no matter how outrageous your claim. Heck, they'll even say stuff like "well maybe the technology has improved so much that he can now throw all those previous rules out the window and tell that someone screaming the in the background is not the person speaking in a 911 tape".

Someone asked earlier to prove he's not an expert. Skeptics asking people to prove a negative . Well here's my proof - he's full of it and he can't say this is not Zimmerman and to claim he can is utterly outrageous. And that's going to be fully revealed by the end of this fiasco.

 

[crimresearch]

more complete CV:

http://www.owlinvestigations.com/pdf/TOM_OWEN_Vitae_01_2011.pdf

BTW - now the report is that Tom Owen claims the easy voice software program is better than 99% accurate (3:00 in video - he says he has tested > 300 voices)

http://video.msnbc.msn.com/msnbc-tv/46919260/#46919260

He has tested. And the rest of the field? How widely accepted is it by the discipline as a whole?

That's the Frye test used in Florida courts, that so many people are playing Pinball Wizard about.

 

[Professor Yaffle]

The thing that is utterly hilarious here is the people insistent on defending Owen either have no clue how easy it is to become an "expert" or don't care one way or the other.

Here's how to become an expert of this type:

1. Take a course for a few days and you're a "voice identification expert".
2. Now if you know what you are doing, you get heavily involved in all of the industry organizations. This gives you cred and you start to get called as an expert witness.
3. Now write papers for organizations that have much, much lower standards that peer reviewed scientific journals, and are easy to get published in.
4. Now start to promote yourself in the media.
5. Do it for many years.

And bingo, you'll have people on JREF swearing your a established expert no matter how outrageous your claim. Heck, they'll even say stuff like "well maybe the technology has improved so much that he can now throw all those previous rules out the window and tell that someone screaming the in the background is not the person speaking in a 911 tape".

Someone asked earlier to prove he's not an expert. I guess we're back to people who purport to be skeptics asking people to prove a negative. Well here's my proof - he's full of it and he can't say this is not Zimmerman and to claim he can is utterly outrageous. And that's going to be fully revealed by the end of this fiasco.

From a quick glance at his CV, it seems he is pretty well regarded in his field.

 

[Xulld]

I just cant listen to that tape, and think to myself other skeptics can also listen to that tape, and then not be skeptical of anyone making claims about matching those distorted, distant, and panicked screams with confidence.

Having read over the techniques only briefly my skepticism is enhanced, not satiated.

 

[Natural Born Skeptic]

OK, now I see. This was the paper it was suggested we all pay $35 for. So again, quoting from a paper we cannot see just doesn't allow anyone to see the context or properly evaluate the context.

I was not referring to the $35 paper. Here is a paper written by 6 scientists.

http://www.afcp-parole.org/doc/AFCP_SpLC_HotTopicsEurospeech03_final.pdf

These are top level people with doctorates in everything from electrical engineering to speech identification. Here are two of them, the others can be looked up too.

http://www.ll.mit.edu/mission/communications/ist/biographies/campbell-bio.html

http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/b/Bonastre:Jean=Fran=ccedil=ois.html

I should note this paper was written in 2003 and I'm sure people will tell me how technology has advanced so much that even though these scientists say voice prints are not reliable even in _much_ less extreme circumstances, suddenly Mr. Owen can do it using a background scream on a 911 call.

 

[Natural Born Skeptic]

From a quick glance at his CV, it seems he is pretty well regarded in his field.

I don't doubt that he is. But I don't accept that his field shares the same level of legitimacy as other fields. I think most of the scientists who are the true experts will never be called, because they wouldn't make good witnesses saying "there is no way to say with any degree of certainty who that is".

 

[Unabogie]

http://abcnews.go.com/US/trayvon-ma...orge-zimmerman/story?id=16055412#.T3o4idVA01I

No broken nose and the wound on his head seems minor?

But Dr. Vidor Friedman, president of the Florida College of Emergency Physicians, remains unconvinced.
"If somebody had been beating his head against concrete I'd think we'd see more obvious scrapes," Friedman said. He also said he would expect to see bandages on Zimmerman's head.
More significantly for Friedman was the condition of Zimmerman's nose.
"All of the ridges in his nose are clearly defined. You would expect significant swelling in the hour or two after a break. There appears to be none. It doesn't look like his nose was broken or badly broken," Friedman said.

So we're back to where we were. The police hopefully took photos which will clear things up.

 

[tyr_13]

In an interview on CNN, Owen said that to testify that the voice in question IS a match, he'd want 60%.

 

[RandFan]

And bingo, you'll have people on JREF swearing your a established expert no matter how outrageous your claim. Heck, they'll even say stuff like "well maybe the technology has improved so much that he can now throw all those previous rules out the window and tell that someone screaming the in the background is not the person speaking in a 911 tape".

Would it not stand to reason that acting inappropriately would put his reputation at risk? What is his motivation to put his years and credibility on the line?

You are creating straw men and begging the question. No one is arguing that Owen is making outrageous claims and is still an expert regardless. And it is reasonable to assume that the technology could advance sufficiently to change the rules. Why is that absurd? Do you deny that the technology of today isn't significantly better than it was 15 years ago? Come on, you are browbeating. It's not helping. Rhetoric has a place but you should rely more on the weight of your argument.

 

[sgtbaker]

http://abcnews.go.com/US/trayvon-ma...orge-zimmerman/story?id=16055412#.T3o4idVA01I

No broken nose and the wound on his head seems minor?



So we're back to where we were. The police hopefully took photos which will clear things up.

A doctor who's never seen him is basing his diagnosis on a grainy video that never gives a close up on his nose. I'll tell you, now I am convinced!

Sarcasm directed at doctor, not at you Unabogie.

 

[RandFan]

I don't doubt that he is. But I don't accept that his field shares the same level of legitimacy as other fields. I think most of the scientists who are the true experts will never be called, because they wouldn't make good witnesses saying "there is no way to say with any degree of certainty who that is".

This doesn't make any sense, the defense cannot or will not call "true" experts that cannot make a compelling case? BTW: what is a "true" expert? Do they put milk on their porridge?

 

[Professor Yaffle]

In an interview on CNN, Owen said that to testify that the voice in question IS a match, he'd want 60%.

He said 60% plus a green light from the software, which also takes into account the other figures like the false rejection figure. As far as i can tell he seemed to be saying that low numbers on all three indicators (I missed what the they were all called as the presenter kept pressing him for a single figure) meant that there was a very low probability of a match.

 

[tyr_13]

He said 60% plus a green light from the software, which also takes into account the other figures like the false rejection figure. As far as i can tell he seemed to be saying that low numbers on all three indicators (I missed what the they were all called as the presenter kept pressing him for a single figure) meant that there was a very low probability of a match.

True. I was just pointing out that focusing on the 48% number was wrong.

I still have reservations about the applicability of this software to this case, but those would be addressed by the courts I would hope.

 

[RandFan]

I still have reservations about the applicability of this software to this case, but those would be addressed by the courts I would hope.

And I have no problem with that. IMO: the expert testimony, unchallenged as it is, only meets the probable cause standard. I would reject the testimony from a "beyond a reasonable doubt" standard as it stands right now. We are not "finders of fact" in a legal sense and an experts testimony alone is not good enough.