Trayvon Martin, Vigilante Justice

[triforcharity]

When did a court of law accept into evidence the claim that a scream could be matched to a spoken voice? I must have missed that link, can you post it again?

They didn't. Can you post a link that any court has rejected a voice analysis of a scream versus a spoken voice?

My point is simple WC. Voice biometric software and analysis is accepted into evidence in many courts of law.

That is my ONLY point. You can continue to intentionally disrupt the forum all you want, but you're simply wasting bandwidth.

 

[Skeptic Ginger]

BTW, I keep seeing it repeatedly stated that courts "accept this type of evidence", correct me if I'm wrong on that. But my research is showing me that it's a huge grey. Sometimes courts allow it, sometimes they don't. Here is a statement from the "expert" Thomas Owen himself, who while arguing about how reliable it is states:

http://www.experts.com/Articles/Voi...-Aural-Spectrographic-Method-By-Thomas-J-Owen

We posted a citation earlier that this kind of evidence was admitted the in vast majority of federal cases. Did you miss that post?

I notice you cherry picked Owen's own words. Here, let me cherry pick mine:

One of the reasons for such opposition regarding admissibility is that the method has evolved greatly since its initial application. Court decisions based on early methods of voice identification analysis are not applicable to the methods used today. No longer are voices compared on the basis of a limited group of key words. Today's aural/spectrographic voice identification method takes advantage of the latest in technological advancements and interweaves several analyses into one procedure to produce an accurate opinion as to the identity of a voice. This modern technique combines the experience of a trained examiner performing the visual analysis of the spectrograms and aural analysis of the recordings with the use of the latest instruments modern technology has to offer, all in a standardized methodology to assure reliability. Court decisions reviewing the early voice identification cases may not be relevant to present day cases because the older decisions were based on less sophisticated procedures. Most of the courts which have rejected admission have been aware of continuing work in this field and have specifically left the door open as to future admissibility.

Here is a book on Google that is someone dated (1979) that is nevertheless interesting simply because it gives insight into how the issue has progressed back and forth over the years up to that time.

http://books.google.com/books?id=Fj... on Evaluation of Sound Spectrograms"&f=false

Why should anyone bother with a 1979 book that would be completely irrelevant to this discussion 40+ years later.

 

[RandFan]

Show me a crockaduck? Don't have a crockaduck? That's what I thought. See, evolution isn't true. Jesus.

 

[WildCat]

I see all kinds of evidence that the software the "expert" used can match a scream to a voice.

All those links are to these same 2 clowns regarding the Martin incident.

 

[totoro]

Screams aside - isn't one of the questions voice biometrics vs voice spectrography ? It seems voice biometrics is much newer and less proven ? Or am I misunderstanding ?

Also, for all the experience, didn't Mr. Owen just run the voices through his software ? it's all inside the magic software. Two voices go in, a result comes out. So - unless you either a) show me the code or b) show me the studies - can't I be a little suspect that maybe he is trying to sell a product ? And that maybe the lack of a match is a failure of the product ?

Also - does this science trump 2 eyewitnesses + zimmermans claim that he was yelling for help ??

Here is a link to an abstract on the standards of the ABRE:
http://expertpages.com/news/american_board_voice_comparison.htm
This standard specifies recommended practices for the handling, preparation and analysis of recorded evidence to be followed by practitioners of the aural/spectrographic method of speaker identification.
If accurate, does it appear that methodology was followed at all ?

You know what would be interesting and useful .... to see if any recordings the parents must have of TM match the call either. Because if neither match, then you got something interesting.

Biometrics is certainly newer. Not sure about less proven: one of the things about that kind of system is that you can get a gauge of its accuracy as you train it.

From the bit of reading I've done, it looks like speaker recognition today is pretty much a branch of machine learning. NIST has been having bakeoffs since at least 2001, and it looks like that's been completely about biometrics. I'm not even sure there is any active academic research on spectrographic analysis.

One of the reasons things have changed is that computers are so powerful and cheap now. The techniques we routinely use on desktops now would have required supercomputers not that long ago, and would have been completely impracticable 30 years ago.

 

[Natural Born Skeptic]

Why should anyone bother with a 1979 book that would be completely irrelevant to this discussion 40+ years later.

Some of us actually like to learn, if you don't, that's OK with me .

 

[WildCat]

The point I am making is that nowhere in the literature have I seen it mentioned that yelling/screaming is a huge problem. Different telephones, yes. Yelling, no.

You made a claim that all the papers in the literature mention that screaming is a problem. I called you on this, and asked for some citations. Since you haven't actually read any of the papers, you were incapable of coming up with any citations to back your claim.

So you come back to your Russell's teapot argument. No I have not read any papers that mention classifying screaming text.

At the same time, I have not heard of it being a huge problem, so I'm not willing to say that it is. Speaker variations due to mood or aging are mentioned, but are always mentioned as a distant second to problems with different microphones or telephones.

As I said, you are just making things up. You claimed that all of the papers in the literature mention this as a problem. When confronted with someone who has actually bothered to read a few papers, you went straight for misdirection.

Tendentiousness combined with an argument from ignorance combined with ad hominems combined with just making **** up. That's what you've given us so far.

How about this? You give us any evidence whatsoever that screaming in a test utterance is a major problem for speaker recognition systems. With your deep understanding of the problem, this should be really easy for you, right?

So you can't find a single paper where a scream can be matched to a spoken voice?

I already explained the relevance - all the voice recognition software/techniques requires the 2 samples repeat the same or similar phrases. The scream and Zimmerman's voice on the 911 calls are not even remotely similar to each other, that is a huge problem when comparing the 2.

You keep ignoring this inconvenient fact.

 

[WildCat]

Screams aside - isn't one of the questions voice biometrics vs voice spectrography ?

It's just 2 terms for the same thing. Both amalyze audio spectrographs.

 

[WildCat]

I think Wildcat will look back on this episode with considerable embarrassment. The way he's handled this issue is truly beneath him.

I get it. All of the people can see why this is damning and they have a hunch that the experts are going to agree that the scream isn't George Zimmerman. I guess if he ever gets put on trial, he'll have an opportunity to challenge all this. Good.

But this hasn't been a shining moment for some of our veteran members.

I'd be shocked if any of this nonsense will be introduced in a trial.

 

[Skeptic Ginger]

Screams aside - isn't one of the questions voice biometrics vs voice spectrography ? It seems voice biometrics is much newer and less proven ? Or am I misunderstanding ?

Also, for all the experience, didn't Mr. Owen just run the voices through his software ? it's all inside the magic software. Two voices go in, a result comes out. So - unless you either a) show me the code or b) show me the studies - can't I be a little suspect that maybe he is trying to sell a product ? And that maybe the lack of a match is a failure of the product ?

What crap. He was asked for his expert opinion and he gave it. I'm guessing he was either paid by the paper or indeed volunteered because there would be publicity involved. Any claims he's only motivated by the desire to sell a product are completely unsubstantiated. In contrast his well documented interest in excellence in the profession as evidenced by his decades long interest in accreditation of the experts and his writings in professionally oriented pieces has been substantiated.

Also - does this science trump 2 eyewitnesses + zimmermans claim that he was yelling for help ??

Unquestionably even without the obvious problem you left out, there are at least 2 additional witnesses who contradicted the ones you are citing.

Here is a link to an abstract on the standards of the ABRE:
http://expertpages.com/news/american_board_voice_comparison.htm
This standard specifies recommended practices for the handling, preparation and analysis of recorded evidence to be followed by practitioners of the aural/spectrographic method of speaker identification.
If accurate, does it appear that methodology was followed at all ?

The expert called on by the paper is an expert in forensics, not just in voice recognition. But he also was not preparing court testimony. Your challenge is not necessarily relevant.

You know what would be interesting and useful .... to see if any recordings the parents must have of TM match the call either. Because if neither match, then you got something interesting.

The only recordings I have of my 22 yr old son are some when he was a child. So "must have"? Don't know what you base that on unless one can get erased phone messages.

 

[Skeptic Ginger]

Show me a crockaduck? Don't have a crockaduck? That's what I thought. See, evolution isn't true. Jesus.

That's so much more communicative than the way I said it.

 

[WildCat]

They didn't. Can you post a link that any court has rejected a voice analysis of a scream versus a spoken voice?

My point is simple WC. Voice biometric software and analysis is accepted into evidence in many courts of law.

That is my ONLY point. You can continue to intentionally disrupt the forum all you want, but you're simply wasting bandwidth.

The problem is this goes far beyond what has been introduced before. It's one thing to compare 2 voice samples repeating the same phrase, quite another when you try to compare one voice sample to a short, low-quality scream.

Again, all of the literature on this compares samples of voices saying the same phrase. What these clowns claim to be able to do is far beyond what has been shown before.

 

[Skeptic Ginger]

Some of us actually like to learn, if you don't, that's OK with me .

????

I am having trouble parsing your support of the 40+ yr old book with your

Forgive me but I'm satire challenged.

 

[bobdroege7]

"can't convict on less than 20 words"

Ed Primeau

"4. Ask for a speech exemplar from George Zimmerman screaming for help"

also Ed Primeau

So I would conclude from that, that one speech recognition expert Ed Primeau can not identify George Zimmerman as the one screaming by comparing the screams on the 911 tapes with the words of George Zimmerman on the other 911 tape.

Edited to add source http://www.audioforensicexpert.com/blog as I am not sure Ed is one of the two experts under discussion.

 

[WildCat]

"can't convict on less than 20 words"

Ed Primeau

"4. Ask for a speech exemplar from George Zimmerman screaming for help"

also Ed Primeau

So I would conclude from that, that one speech recognition expert Ed Primeau can not identify George Zimmerman as the one screaming by comparing the screams on the 911 tapes with the words of George Zimmerman on the other 911 tape.

HERETIC! Crockaduck!!111eleventy1!!!!!!!!

 

[Natural Born Skeptic]

For those who are truly interested, the paper published here, by a real scientist (as opposed to a guy with a degree in history that took some side courses) who's life is devoted to the study of sound and speech issues and who has been published and cited repeatedly in scientific journals is quite good if you are willing to spend $35.

If you don't want to spend $35 you can find numerous papers he's written on everything from how our vocal chords developed over time to this paper titled Person Authentication by Voice, a Need for Caution.

 

[Natural Born Skeptic]

????

I am having trouble parsing your support of the 40+ yr old book with your

Forgive me but I'm satire challenged.

It's quite simple. I'm interested in the subject and how the courts have viewed it over the years. I find it rather ridiculous when a person posts something to share with others that might have the same intellectual curiosity about a subject and makes an effort to point out the limitations of it as I did, and then someone still has to jump in and say "ah ha, but it's 40 years old". Um ya, I pointed that out when I posted it and merely said it was interesting. So as I said, if you have no interest in learning about it, don't read it.

 

[Skeptic Ginger]

Cases cited on Owen's website (I believe crimresearch asked earlier):

CRIMINAL LAW - VOICEPRINT EVIDENCE IS BEING HEARD WORLEY V STATE, 263 SO. 2D 613 (4TH D.C.A. FLA. 1972)

IN THIS CASE, THE COURT HELD THAT VOICEPRINTS WERE PROPERLY ADMITTED TO CORROBORATE A DEFENDANT'S IDENTIFICATION WHERE SUFFICIENT EVIDENCE WAS AVAILABLE TO SUSTAIN HIS CONVICTION WITHOUT THE USE OF VOICEPRINTS.

ALEA v. STATE (FL)

The state argues that admissibility of voice identification has long been the law in Florida. Such testimony is admissible to establish the identity of a suspect as direct and positive proof, although its probative value is a question for the jury. In Martin v. State, 100 Fla. 16, 129 So. 112 (1930), the Supreme Court of Florida stated the rule as follows:
"This court has held that testimony is admissible to establish the identity of an accused even by one having heard his voice, and such evidence is not to be considered as circumstantial, but as direct and positive proof of a fact, and its probative value is a question for the jury. Pennington v. State, 91 Fla. 446, 107 So. 331; Mack v. State, 54 Fla. 55, 44 So. 706, 13 L.R.A.,N.S., 373, 14 Ann. Cas. 78; Martin v. State, 86 Fla. 616, 98 So. 827."

More cases are cited in that last link.

 

[totoro]

So you can't find a single paper where a scream can be matched to a spoken voice?

I already explained the relevance - all the voice recognition software/techniques requires the 2 samples repeat the same or similar phrases. The scream and Zimmerman's voice on the 911 calls are not even remotely similar to each other, that is a huge problem when comparing the 2.

You keep ignoring this inconvenient fact.

Nothing I've read says this, as long as the samples are long enough. If you'd bothered to actually read any of the papers instead of just pretending you had, you would know this.

 

[Skeptic Ginger]

It's quite simple. I'm interested in the subject and how the courts have viewed it over the years. I find it rather ridiculous when a person posts something and makes an effort to point out the limitations of it as I did, and then someone still has to jump on and say "ah ha, but it's 40 years old". Um ya, I pointed that out when I posted it and merely said it was interesting. So as I said, if you have no interest in learning about it, don't read it. I pointed out its limitation when I posted it.

Right.

And just what do you expect to learn about current technology from 40+ yr old textbook?