NervousXtian
Thought Emoji Movie was good. Take that as you will.
Doubtful he did, but I did and it answered some questions about the concerns I first had of the story.
Family of Florida boy killed by Neighborhood Watch seeks arrest
- Thread starter Dram
- Start date
- Status
Doubtful he did, but I did and it answered some questions about the concerns I first had of the story.
ViperVisor
Member
It is just hard to apply that in a case where he shot and killed someone as a matter of fact. The justice question is how did it happen.
demigod
Member
I can recognize my nephews regular voice and when they scream. Zimmerman's voice and the screaming is COMPLETELY different.
While I'm not versed at all in voice identification, I'm perplexed that you would come to this conclusion based on the figures cited. This reeks of scientific illiteracy on your part, as well as a lack of common sense -- if the standard for a positive match by this metric is 90%, 48% quite clearly suggests that it is VERY UNLIKELY for it to be the same voice, and not a 50/50 chance as you believe. As others have pointed out, they are both human voices, both male, and, as I would imagine this probably has some effect on voice development, they are both native English speakers as well. It would be reasonable to presume that these factors combine to account for the 48% match.
As an analogy, though again I don't know much about voice identification at all, I would liken it to homologous sequences in animal genomes. There is a high incidence of homologous DNA sequences between vastly different animals, and this is part of the overwhelming network of evidence for evolution. However, if we apply your logic in this context, since a fruit fly shares 60% of its DNA with humans, there's a 60% chance that George Zimmerman is actually a fruit fly. Those "experts" and the way they spin their facts!
In addition, Owen has a website for his consulting business that includes some articles written on forensics, and specifically what his requirements are for what he believes is a valid evaluation. He goes over the need for multiple samples where the same words are used and, most damningly, the need to have the delivery be similar between the exemplars and the audio to be identified.
www.owlinvestigations.com/article2.html
www.owlinvestigations.com/article2.html
It's not quite as damningly as you put it. Definitively, in a controlled trial they would want to have a screaming voice to compare to a screaming voice, but that is not to say that they couldn't use other recording to identify and reasonably omit a sample. He's stating the optimal settings for voice identification as forensic evidence in a court of law. To give the defense and prosecution as little possibility as he can to question the findings, because of course if there's any possibility of doubt, they would play for it. Apparently software can still find voice print data regardless of the volume or pitch of the voice to match a target from a sample.
http://www.youtube.com/watch?v=xpgF9-gILPY
http://www.youtube.com/watch?v=xpgF9-gILPY
That marketing video did not substantiate what you're claiming. Where is the differing pitch? Where are the different deliveries?
If you made it past that awesome 90's rock intro, he has a normal speaking voice, a sample where he disguises his voice with a lower pitch, and other samples where the voice is obscured or degraded. The software still found all samples to be the same person within a <90% chance of accuracy.
While I certainly see the need for forensic voice identification in the courtroom, I don't get the recent fascination with it in this thread. Disregarding subjective judgments regarding how old the person screaming for help sounds like (and by 17 I'd expect that one would have an adult voice anyhow), what possible situation would have Zimmerman screaming for help when he's the one holding the gun and then stop screaming after he shoots Trayvon? And how would it be possible to have uninterrupted screams such as those recorded in the 911 call if you're getting your head pounded against the pavement?
But yeah, I guess the fact that people feel the need to speculate on such inane matters speaks volumes about how little actual progress there is in the investigation.
But yeah, I guess the fact that people feel the need to speculate on such inane matters speaks volumes about how little actual progress there is in the investigation.
Unknown One
Member
What is Zimmerman? A voice actor? The tone and the way he sounds is totally different from the screaming and it makes no damn logical sense. The screams stop immediately after the shot. Who the hell screams before shooting a kid? Sounds like a looney to me.
But the sample in this case is not degraded or obscured, nor is it simply a lower pitch. The man outlines very specifically what he needs and all of those basic requirements remain unfulfilled in the case of the Zimmerman audio. And, again, this is a marketing video designed to sell a product. We cannot seriously take it as a valid representation of the software's performance.
demigod
Member
lol so typical of you. The guy even disguised his voice, what more do you want?
I've said that before, it makes ZERO sense for him to stop screaming for help after the gunshot went off. And yeah you're right, if he's being pounded to the pavement, the scream would be totally different. The pitch would be different when he's yelling out each word.
That kind of logic has no place here in this place totally devoid of all the facts. We have to wait. Wait and see. It's possible that he was screaming for someone to save him from having to shot this kid, and then after having shot him, any more screaming would be moot.
Harv, I know that's an ad, dude, you don't have to keep saying it. I know you would prefer a video from a professor of acoustic engineering or something, but we don't got any. The other guy's software says there's a 48% match, meaning a very low chance of it being Zimma Zimma doing the screaming, so the differences in pitch did not appear to be coming from the same person. Maybe they need to have Vigilante Zim try to match as best he can, for another go at testing. If he's innocent, his lawyers should be asking for the chance to do it.
edit:
here's me talking normally. and now here's me sounding like sam elliot from the big lebowski.
do you have to be so obstinate, dude?
Disguised it how? And was it the same audio clip just modified with things like music or lowered pitch? I am a bit surprised there is any desire here to attempt to defend this analysis. If you think this caliber of evidence is a good thing for your desired outcome you really haven't thought this through. Weeding out this nonsense serves everyone's interests.
That video shows the software that was used and the guy using it(and the expert quoted in the media!) not only make the software but wrote the paper I linked to.
huh? I think it's different software and different dudes, dude.
demigod
Member
Go buy the software and prove it to me that it's not accurate, otherwise I will believe it over your *opinion*.
Nope, same software sold by the same guy.
This. If you're skeptical of the validity of this software due to your expertise, then maybe you can recommend better software to the NSA, CIA, and FBI, because clearly they shouldn't trust this software either. It's not like they make tracking terrorist targets via telephone a high priority or anything.
Unknown One
Member
The only way I'll believe him is if he actually replicates the screaming which I doubt he will. The defense will just bring up their own voice analysis expert in court. His story doesn't add up at all, there is no reason for me to believe in him until there is actual facts backing up his and his father's statements.
Well the software in the video is EVB - Easy Voice Biometrics, and the dude Thomas Owen the speech recognition guy from the news uses a number of different software packages, Kay Pentax, Ikar Voice Analysis, and others according to his website. I don't think the dude in the video is Thomas Owen.
So... as I posted before, a 48% match does not mean that there is a 48% chance of it being Zimmerman's voice. Let's try a different analogy here: if we compare two paintings, there will certainly be similarities in color, texture, and shape. If they are similar subjects or employ similar color schemes, then by an objective measure they may share 48% of these elements in common. Obviously, though, there is NOT a 48% chance of them having been created by the same painter. Going further with this analogy: if one were to analyze two paintings on a hunch that one is an unknown painting that should actually be attributed to a famous painter, you would certainly need more than 48% similarity in these aspects between the unidentified painting and the current sample of the known painter's work, along with testing it with other works of the painter. A 90% similarity across the board seems like it might be a reasonable standard to say that it is probably by the same person.
As far as the first part of your post, that's asinine to me and I think it would be prudent to apply Occam's Razor here. Which do you think is more likely, Zimmerman screaming for help (when it's him holding the gun) and then NOT screaming for someone to call 911 after pulling the trigger, or Trayvon screaming for help because someone's holding him at gunpoint until he is no longer physically able to scream for help?
Why is it awful? The media has been terrible. Zimmerman was a white 240 lb racist with no injuries on his head that called Trayvon a coon. None of that was accurate. You have a guy we know stalked an unarmed kid and killed him, isn't that enough without making stuff up to sensationalize the story and fit a narrative?
I will admit the title is a bit of a troll though.
False, that so call enhance video made is garbage,as soon as he starts moving the marks disappear.Also the other facts as the ambulance being canceled and him begin treaty so quickly shows how severe is injures was.
Also the media didn't cherry pick anything, they show the pictures they were giving when this first broke out.
All huffington post doing is feeding the fire of stupid with there own.
In the articles he specifically mentions the EVB software as being what he used. If you type Easy Voice Biometrics in google you get news results.
Ok, so we agree. What's the question again? Ohh..Oh. Sorry man. I was being sarcastic dude.
I'm not sure. I'd like to have it explicitly stated that that was what he used. The news articles I have read after googling seem to be confused whether Easy Voice Biometrics is the process of using computer software for voice identification, or if that was the name of the software that he used. And it seems to be the reporters saying that he used EVB, not Tom himself. So, the jury is out on that one. Or did you see a quote?
I agree that the basic concrete facts in this case should be enough for this to be a terrible, terrible story of injustice without sensationalizing it further. But why are you so quick to denounce the media reports so far as inaccurate? The grainy video released so far is insufficient to prove either the presence or absence of head injuries, though I would say that they are enough to call into question the extent of any injuries as self-described by Zimmerman. The utterance under the breath by Zimmerman is still that (i.e. mostly inaudible), and I don't see how the new "enhanced" audio can really clear it up either way -- it doesn't sound any clearer to me than the original 911 call, and I still hear him saying "fucking coon" as clear as people are claiming they hear "punk," unless it's pronounced poonk in other parts of the country.
No, I read what experts say. The man wrote this all down, you can read it for yourself. He didn't even follow his own prescribed procedures.
Ok, so we agree. What's the question again? Ohh..Oh. Sorry man. I was being sarcastic dude.
Ah, I thought so given your initial post about a bet, but you'll forgive me as it's kind of hard to tell with this thread? :x
A still image is not enhanced video.
Almost none of that is accurate now but based on the info they had at the time it 'was' accurate as far as the world knew at the time. Furthermore the 'coons' crap and punks crap is clearly open to interpretation at this point, you hear what you want to hear as experts, casual observers and even 'friends' of Zimmerman have attributed it could go either way (In fact the Zimmerman camp seems to the be the only one that hasn't denied that he could have said coons, he has a friend almost assuming he did and tries to make it sound like a term of endearment). The weight issue is because they didn't have any new relevant info that wasn't 3-4 years back, same could be said of Trayvon Martin, the only thing else they had were pictures, and that's it. Also I still don't see any damn injuries on his head so how is that false?
This includes some information regarding the FBI's view of audio forensics:
http://www.talkleft.com/story/2012/4/3/03733/45115
Well if they hire him to be a court expert, or if he is called to testify and gets access to motherfucker Zim screaming his bitch ass off, then maybe he can apply his prescribed procedures. But right now, he's using the standard settings, which doesn't invalidate his findings, it just makes them less than perfect. Boosh.
Are you joking? He can't do it right, so he did it incorrectly and we should just use it? How is that number he claims to quantify the amount it "matches" not become totally invalid when he doesn't do the work properly? You're being ridiculous.
What I wish they had cleared up, is whether there's a word before "fucking", and what it may have been.
No, he can't do it perfectly. That doesn't mean that he can't do it at all. That demo shows the software can work when the samples are high and or low, with different pitches, etc. So while he says when collecting samples for a court case he would want to be as clinical as possible, gathering samples to be as similar as they can be, to negate any possible arguments from lawyers about the dissimilar sounds not being relatable, the software, and potentially other processes using other voice identification software can work properly in imperfect conditions. Or at least enough to find 48% similarities which aren't enough to say that the samples came from the same person.
And thank you. I enjoy funny hats for a reason. We're not using anything, this isn't a court of law and you're not a lawyer. It is enough to cast doubt and add to the conversation. If this was to go to court, I would fully expect him to get a legitimate sample to either tie those screams to Invader Zim, or indicate that they were not from him. Which should do wonders for bolstering his self-defense defence. So surely he would be all for it. And you as well, I take it.
His own words and the reports written by the FBI and the forensics boards all make it a point to show how the conclusions drawn using these methods are unreliable unless care is taken. He asks for multiple samples. A case he was an expert in in 1998 used 6 samples. You claim this is only done to satisfy lawyers and judges, yet everything suggests this is required to draw any valid conclusion at all. You watched a marketing video where a product performed well. You told me I didn't need to repeat this but clearly I do, as you seem to think it proves something.
His number means nothing. You want it to, and that's too bad.
ooohh, knifey spoony, eh? that fbi symposium said they wanted to do away with uncertainty, and I should hope so, when attempting to get warrants to do law enforcement actions and such. The same criteria should apply to that as with a court of law. This video however is just a demonstration, and Thomas coming forward was a demonstration and a statement of his findings. Now if we wanted a definitive answer they can get the goddamn multiple samples you go on about and exonerate the little pissfuck, if that's what you want so bad Harvey. So why are we still arguing?
I think a lot of the discussion over how reliable the voice software is is moot. Several witnesses mentioned how the screams sounded like that of a boy or someone young. They'd testify to that in court and even Martin's mom might. That plus the audio experts should be more than sufficient for a jury to conclude that it was Martin screaming for help. Still, this stuff is secondary compared to the forensics and the autopsy reports.
It's a meaningless demonstration given the capabilities of the method being used. You're desperately trying to hang on to the concept that it's "good enough" for the purposes of conversation, but you can't support that claim. What confidence level do we have in his number of 48 if he does it wrong?
I don't know why we're still arguing. There is nothing for you to argue.
You keep claiming he did it wrong, when that's not necessarily true. He didn't use optimal conditions that would meet court or warrant requirements, but that doesn't mean his testing was below the operational standards of the software or voice ID procedures themselves. It's entirely possible that his tests were 'good enough' for proof, just not the definitive proof that would be unquestionable in a court of law.
I know why you're still arguing. It's because you like to argue.
So the symposium, the FBI and this expert pointing out the importance of matching whispers with whispers, stressed voices with stressed voices and identical words are all just being super careful and accounting for skeptical juries? Please. If that process was not required to demonstrate statistically significant, scientifically valid results they would not go through the immense trouble of doing it. They would justify their method and provide their analysis. The method they use, which does not even satisfy the FBIs own standards, calls for these specific steps to be taken.
I am merely providing others with the views of these forensic experts. You are applying your own analysis and determining what is valid or good enough. You can't justify this.
I am not getting anywhere here.
All the things listed go against Zimmerman, I have a hard time putting all of it in the coincidental category for that reason. White, 240, no injury, coon, him in a mugshot, Trayvon at 13 years old. I also left out the 46 or whatever it was calls in a year turned out to be over 7 years or something. Oh and the calling in a 7 year old black kid turned out to be because he was worried for his safety, that was another good one.
OP in this thread certainly had me thinking racist too, facts dont really bear it out yet.
To be fair the family lawyer plays the race angle first, not the media, but the list of "facts" that were wrong is undeniable and a lot of media played it up. This in a case where we know Zimmerman chased an unarmed kid and we know for a fact he killed him, he admits it. All this other BS is clouding it.
soundahfekz
Member
I still have the misfortune of seeing Kharvey's posts through other people's quotes.
Every. Fucking. Contrarian. Position. Every. Time. Now it's about the validity of voice biometrics.
Next we'll hear about how 99.9 percent certainty that it ISNT Zimmerman isn't good enough either. Shits ridiculous, just posting to be annoying as if no one knows that any biometric test done would have to be as scientifically absolute as possible. There is no other motivation for some of the people in this thread other than to be a dick that gets off on sniffing their own shit.
Every. Fucking. Contrarian. Position. Every. Time. Now it's about the validity of voice biometrics.
Next we'll hear about how 99.9 percent certainty that it ISNT Zimmerman isn't good enough either. Shits ridiculous, just posting to be annoying as if no one knows that any biometric test done would have to be as scientifically absolute as possible. There is no other motivation for some of the people in this thread other than to be a dick that gets off on sniffing their own shit.
NervousXtian
Thought Emoji Movie was good. Take that as you will.
Not talking about the enhanced image, but the stupid "enhanced" video of when he got to the police station. It's not more damning of an injury then the pretty crap video was of proving there was no injury.
The only thing to prove it either way is actual pictures, and we don't have them. The police reports are questionable, we all know this.
Makes perfect sense he'd stop screaming if it was Zimmerman, he just shot someone. He'd be in shock. Also on the 2nd part, I've seen someone getting beaten before and they were yelling/screaming the whole time. It's not a pretty sight, and it didn't seem to effect their screams that they were being kicked while lying down.
I can't wait for the real evidence to come out, so we can focus on that and see Zimmerman go to trial.
- Status