Trayvon Martin, Vigilante Justice

[WildCat]

Are you trying that hard to miss my point? Spamming is rude and results in people ignoring your posts. So no, I didn't stamp my feet. I just ignored what he wrote until he tries to engage in a way that's conducive to discussion.

So you're stamping your feet and crying about form over substance.

It's obvious you have no interest in hearing any evidence that interferes with your pre-determined conclusion.

 

[WildCat]

That's if you are using spectrographic analysis. Not at all relevant if you are using biometrics, which do not, in any way, involve looking at visual representations of the samples.

And the difference between spectrographic analysis and biometric analysis is...?

Isn't "biometric analysis" just another way of saying you're analyzing the spectrograph? If it's not analyzing the spectrograph, what exactly is it analyzing? What other graph is generated by sound waves?

 

[sgtbaker]

It would be obvious if you didn't have DNA to sample. But with sound, any sound will do. We already know that for 2 sound samples to be compared they have to be similar, reciting "Mary had a little lamb" for example. This guy compared a scream for help to a normal speaking voice not saying the word "help" in any way, shape, or form. Of course they won't match! They should never have been compared in the first place, but Owen used them anyway. The same Owen who is on record as saying the comparisons have to be of the same phrase.

I am reluctant about the validity of the comparisons, as well, but I thought with biometrics, you only needed voice samples, that don't have to be the same word. I think he said that in the 'how to' video. It was the spectro... something or other that needed identical words. I could be wrong, like I said, this technology is lost on me.

 

[Skeptic Ginger]

I've become a lot more uncertain of even long established identification techniques where expert judgement is required (like fingerprinting) since the Shirley McKie case http://en.wikipedia.org/wiki/Shirley_McKie

But again, it is usually much more simple to rule out a suspect than it is to positively identify one.

Some other fingerprinting errors here http://en.wikipedia.org/wiki/Fingerprint#Instances_of_error

Note that all involve positively identifying someone (incorrectly) rather than incorrectly ruling someone out.

The discrediting of bite evidence is another example.

But I think we are forgetting that one eye witness said the voice was a 'boy's' not a man's voice and Trayvon's parents despite the claim the father did not at first recognize the voice have both said they recognize the voice as their son's. You have kids. Do you recognize your children on the phone and on voice messages?

So, we have GZ's story that is only corroborated by a grainy view of a head abrasion, a police report his nose and head had blood on them and one witness who is sure GZ was being pummeled by TM based on "red clothes".

Against that we have the words on GZ's 911 call, TM's girlfriend's testimony, the timing of that call, the timing of the shooting and the location of the shooting all of which are not consistent with GZ's story. The extent of GZ's injuries are not consistent with GZ's story. The opinions of the voice expert, the parents and one witness is that the call for help was either not GZ, was TM or was 'a boy' respectively.

It's not like the voice analysis is all we have.

 

[Skeptic Ginger]

This is an important point -- just because a technique requires subjective judgment doesn't mean it's not valid. As you correctly point out, many perfectly valid techniques require expert analysis for interpretation.

But, to the extent that a match or mismatch is judged based on an expert looking at a spectrogram and listening to a recording, I want blind testing to confirm that the expert is usually right.

The problem is we only know we haven't found that software validation online. We don't know it does not exist. And what bothers me are all the attempts to discredit Owen when there is so much evidence of his scientific standards and integrity that has been cited.

 

[Skeptic Ginger]

I just find it rude when people spam the thread. If he found something of relevance, he should post a single comment and explain why he thinks it's important. Spamming just comes off as childish and if he has a point, no one will notice it.

It was also deceptive.

 

[totoro]

So you're stamping your feet and crying about form over substance.

It's obvious you have no interest in hearing any evidence that interferes with your pre-determined conclusion.

I think I understand where the complete disconnect between the two of us was. I've been reading fairly current papers in the computer science literature on the topic, which are all about using machine learning to solve this problem, and none of which mention these issues as being hugely problematic.

I think perhaps you've been looking at papers and tutorials which use the approach Owens does. This stuff doesn't seem to have a lot of active research going on, and certainly isn't what the government is funding on a large scale.

So in some sense we've both been right. The question remains about whether the machine learning approach is good enough to supersede the visual inspection of spectrographs.

As I said before, nobody who has posted so far in this thread is actually qualified to make that decision. I'm not sure exactly who that person should be, and whether it should be either of the people being cited.

 

[Skeptic Ginger]

So you're just going to stamp your feet and complain about form rather than address the substance?

So you are just going to stamp your feet and ignore the fact out of context quoting was the substance?

 

[WildCat]

It's not like the voice analysis is all we have.

True, we have 2 eyewitnesses who told police it was Zimmerman who was screaming for help.

 

[Information Analyst]

First I want to call out the loaded statement, "his supporters." Wanting to know all of the availably details in the situation before condemning a person does not equate to supporting the person. Secondly, "it seems most likely," is not fact. That's us playing internet detective.

We know Martin was return to the development, and that Zimmerman was leaving it. Given the obvious routes of both in relation to where they both lived/were staying, the area around the main entrance is the most likely place Zimmerman could have first seen Martin. You are, of course, free to offer up alternative routes for both and a location for that first "meeting" that you think more likely.

A) why is one more likely than the other and b) you are still guessing.

It's more likely to be RVC, because it would explain the <=2m 21s between the end of Zimmerman's call to the police, and the end of Martin's call to his girlfriend, i.e. looking for someone who wasn't actually there, but who he thought was in the area, heading for the back entrance. You are, of course, free to offer up an alternative explanation for the missing <=2m 21s.

I would imagine a few lights went on when the fighting began and more went on when the shot rang out. Porch lights being one when there is a police investigation in your back yard is not evidence that they were on before the circus started.

Others have suggested that the area would be unlikely to be unlit under normal circumstances, whether by house lights or communal lighting. Given that Zimmerman expressed concerns about burglaries in the area, if seems rather unlikely that the residents could think that such an area being unlit at night would be a good thing.

Everything about this is just wrong. Those you accuse of "supporting Zimmerman," are not supporting Zimmerman, they are supporting the finding of facts. Don't project your need to be right on other people.

This whole forum is a exercise is people offering their interpretation of various events/situations/etc. The fact that you seem so uncomfortable about me and others offering ours kind of proves my point.

I asked for facts and got what you think happened based on your conclusion that he is guilty.

No, I'm looking at what evidence we know of with the benefit of detached objectivity, living as I do in a country in which this particular situation can effectively never be replicated. Whatever happened in this case has simply zero impact on where I live, and it doesn't affect my lifestyle one iota. It doesn't challenge the idea that I can possess something that I can't possess anyway, nor put me in fear if it sanctions the same possession (which doesn't exist) by other people.

I actually consider that what I think is likely to happened is actually far more lenient towards Zimmerman than you seem to think (e.g. it actually validates his "returning to vehicle" claim), especially since I didn't directly address what happened after the end of Martin's phone call.

 

[Skeptic Ginger]

Here are more excerpts from that first paper (2009):

[Page 2]: "In general, phonetic variability represents one adverse factor to accuracy in text-independent speaker recognition. Changes in the acoustic environment and technical factors (transducer, channel), as well as
“within-speaker” variation of the speaker him/herself (state of health, mood, aging) represent other undesirable factors. In general, any variation between two recordings of the same speaker is known as session variability [111, 231]. Session variability is often described as mismatched training and test conditions, and it remains to be the most challenging problem in speaker recognition. "

[Page 16]: "Any variation in different utterances of the same speaker, as characterized by their supervectors – be it due to different handsets, environments, or phonetic content – is harmful."


I encourage everyone to read the entire paper, but certainly screaming versus spoken would qualify as severe "within-speaker" variation and have a corresponding effect on accuracy.

Again with ignoring the point of the paper which was to point out all the problems and then NOTE THAT NEWER TECHNOLOGIES OVERCOME THESE LIMITATIONS.

 

[Information Analyst]

The news referred to his automobile as an SUV, in the beginning, then it became a truck. I have no idea.

Zimmerman himself describes it as a truck.

 

[Moss]

Again with ignoring the point of the paper which was to point out all the problems and then NOTE THAT NEWER TECHNOLOGIES OVERCOME THESE LIMITATIONS.

Though as totoro noted we have no idea if Owens actually uses the newer approach.

 

[johnny karate]

Shocking. All of the internet doctors already informed us there was no visible evidence.

And based on the level of skepticism being directed at the audio analysis we should accept this new "evidence" because... ?

 

[totoro]

And the difference between spectrographic analysis and biometric analysis is...?

Isn't "biometric analysis" just another way of saying you're analyzing the spectrograph? If it's not analyzing the spectrograph, what exactly is it analyzing? What other graph is generated by sound waves?

I agree that "biometric analysis" is a marketing term. But the spectrographic analysis that Owens is talking about involves people looking at visual representations of speech data.

"Biometric analysis" (or whatever you want to call it) involves using automated procedures to take that data, convert it into feature vectors, and run it through classifiers or something similar to decide whether a test instance belongs to the class of "same as" the training data. No visual representations are involved.

What kinds of distortion would perturb such a system isn't something I would want to speculate on, but what the papers seem to be most concerned about is differences in phones and microphones, which skew the whole dataset in a systematic way.

Without knowing how the incoming data is converted into vectors, what the dimensions in the vectors represent, what kind of smoothing and preprocessing is done, how many dimensions there are, how many vectors are formed from a particular sample, how sparse the dimensions are, what the distribution of values is across dimensions, what kind of model is being used, and a bunch of other stuff, I wouldn't even want to speculate about this.

There is no reason to assume that data issues that are problematic for visual inspection of spectrograms are also problematic for machine learning approaches. Conversely, there may be data that are hugely problematic for the machine learning approaches that are not problematic for the visual inspection approach.

I'm an expert in machine learning (in another domain, but the techniques are still very similar), and I wouldn't want to say one way or the other.

 

[Skeptic Ginger]

And the difference between spectrographic analysis and biometric analysis is...?

Isn't "biometric analysis" just another way of saying you're analyzing the spectrograph? If it's not analyzing the spectrograph, what exactly is it analyzing? What other graph is generated by sound waves?

Based purely on the definitions of the words, the older spectrographic analyses would look purely at the audio analysis.

Adding a biometric component one would match certain sounds to the shape of the throat and larynx. So one could then compare different sounds.

Say for example throat/mouth/larynx A results in a range of sounds and you can make a model of those sounds. You also take throat/mouth/larynx B and make a similar map.

Now you compare the two and you see where they overlap and where they are distinctly different.

Then you do the same with a very large sample of throat/mouth/larynx recordings.

What you come out with in the end is a pattern that reflects the physical differences between individual anatomy reflected on an audio wave screen. Just like fingerprints don't rely on matching one print to another which with 6 billion people on the planet would be useless, different parts of the print are categorized. You can then eliminate various prints as not matching right away because they don't have certain characteristics. You don't have to match the whole print.

TM and GZ have different shaped throats/mouths/larynxes. If the voice identification/recognition software has identified certain key points in an audio recording that are recognizable as matching or not matching a certain biometric then spectrographic and biometric techniques would indeed be very different. The former would rely on simple matching while the latter would rely on matching the person's anatomical differences as reflected by the spectrographic print out.

 

[WildCat]

Zimmerman himself describes it as a truck.

People often refer to SUVs as trucks.

 

[Cylinder]

I just want to know where Mr Owen got the 40+ points of comparison (provided that the media reports of 48% match are true) from the snippet of questioned audio the he claims would be required for eliminating Zimmerman as the donor - or even the 20 points he would need for a positive elimination assuming a 0% match rate.

 

[sgtbaker]

We know Martin was return tpo the development, and that Zimmerman was leaving it. Given the obviously routes of both in relation to where they both lived/were staying, the area around the main entrance is the most likely place Zimmerman could have first seen Martin. You are, of course, free to offer up alternative routes for both that you think more likely.

It's not a matter of me thinking something is more likely than the next. I am saying that I have consistently watched people fill in what they don't know with details that match the conclusion that they've already reached, up to and including Zimmerman's intents.

It's more likely to be RVC, because it would explain the <=2m 21s between the end of Zimmerman's call to the police, and the end of Martin's call to his girlfriend, i.e. looking for someone who wasn't actually there, but who he thought was in the area, heading for the back entrance. You are, of course, free to offer up an alternative explanation for the missing <=2m 21s.

Or, it's possible that he stood there for a while, contemplating whether or not he should pursue or go back to his car. Or he could have walked towards where Trayvon was last seen, and then figured he got away, too. Or he could have said, "damnit, there is no way this ******* is getting away," and ran after him. You don't know and you are deciding that this scenerio is more likely because you've already reached conclusions about Zimmerman.

Others have suggested that the area would be unlikely to be unlit under normal circumstances, whether by house lights or communal lighting. Given that Zimmerman expressed concerns about burglaries in the area, if seems rather unlikely that the residents could think that such an area being unlit at night would be a good thing.

Or, people could be turning off any unused lights because it's a bad economy (Orlando reportedly being one of the worst in the nation), and they leave lights off to keep costs down. Or, being I live about 60 miles away, they keep their lights off for the same reason I do. Florida is mosquito infested, especially considering the exceptionally warm winter that didn't touch the population. Your lights on ensures that you will have an ass load of bugs on your window ranging from gnats, to mosquitos, to palmetto bugs (giant flying roaches), just waiting for you to open the door. If I so much as leave the curtain open, the light from my bedroom attracts every mosquito in the neighborhood. Again, you create a scenerio where everyone must have left their lights on, because that way, you can safely say that there is no way Zimmerman actually lost sight of Martin because that doesn't fit your conclusion.

This whole forum is a exercise is people offering their interpretation of various events/situations/etc. The fact that you seem so uncomfortable about me and others offering ours kind of proves my point.

The forum also consists of people challenging interpretations.

No, I'm looking at what evidence we know of with the benefit of detached objectivity, living as I do in a country in which this particular situation can effectively never be replicated. Whatever happened in this case has simply zero impact on where I live, and it doesn't affect my lifestyle one iota. It doesn't challenge the idea that I can possess something that I can't possess anyway, nor put me in fear if it sanctions the same possession (which doesn't exist) by other people.

I actually consider that what I think is likely to happened is actually far more lenient towards Zimmerman than you seem to think, especially since I didn't directly address what happened after the end of Martin's phone call

It's not a matter of being lenient to Zimmerman. If you haven't recognized that by now, there is nothing more I can say.

 

[totoro]

I just want to know where Mr Owen got the 40+ points of comparison (provided that the media reports of 48% match are true) from the snippet of questioned audio the he claims would be required for eliminating Zimmerman as the donor.

The stuff Owen is doing really does seem as much art as science. That's one of the things that makes the machine learning approach seem attractive: there is still some art in the preprocessing, but overall it's a lot easier to quantify. I would imagine that the intelligence community also likes the idea of looking at a bunch of phone calls in an automated way to see if they can find a particular person's voice, which wouldn't be practical on Owen's approach.

The big open question none of use knows the answer to is exactly how good the machine learning stuff is and whether it's admissible in court. It looks like at least some software based on this approach has been ruled to be.