Voice ID experts

[MJW]

I don't know if there are any cases where that error is held to be harmless.

Yes, for example, Flanagan v. State, 625 So.2d 827 (Fla. 1993). I can't see how, in a weak case where the identity of the screamer is critical, allowing in improper voice ID evidence could be held harmless beyond a reasonable doubt.

[Cylinder]

But my point was a different one, and it went to how I speculate Nelson will rationalize allowing the testimony.

If the court chooses to adopt the state's view, she will rationalize it, IMO, by name. Instead of looking under the hood she will just assert that voice recognition is accepted and call the details an issue of fact instead of law. As you pointed out on another subject, the state uses a simple strategy - everything they assert in their response is right. Differing opinion or claims of experimental error aren't Frye territory. Of course, none of that is helpful because Owen and Reich aren't following an accepted methodology. It would be like doing DNA matching with a dowsing rod. It's DNA matching, right? We all agree that's an accepted science.

I was pleased to see that MOM was measured in approach to the H&H report. What they did was ethical science. It's just not reliable and they have the integrity to point that out in their own report. That should be rewarded.

[cboldt]

Yes, for example, Flanagan v. State, 625 So.2d 827 (Fla. 1993). I can't see how, in a weak case where the identity of the screamer is critical, allowing in improper voice ID evidence could be held harmless beyond a reasonable doubt.

Neither can I.  Well, except to the extent the experts can't reach a conclusion - and Reich is about as good as an astrologer, (tongue in cheek) I can see the DCA ruling that no reasonable jury would be persuaded to adopt the conclusions of the experts, based on the expert's testimony.

But seriously, I agree with you, allowing the experts would be reversible error.  I think Nelson doesn't care.  Either because she wants to pass the buck to the DCA, or because she thinks the defendant is going to prevail.

[RickyJim]

I predicted some time ago that she'd have a hearing, then let the experts testify.  The basis will be that the jury is able to sort out an conflicting conclusions.  The state's case, without voice experts, is exceptionally weak.  By now, Nelson probably has a sense of just how vulnerable the state's case is, and she'll be throwing lifelines as quick as Bernardo gives half a reason to do so.

Really?  If in a televised case she is making ridiculous decisions, it won't adversely effect her career?  I am not a Frye expert but I doubt, "Let's just let the jury handle it." is an option for her.  She has to rule whether or not the expert is applying accepted science.  A judge who allows an astrologer to testify might well be in trouble.  Bernie's only argument is that his experts use spectral analysis so they are OK; let the details of how be damned.  Astrologers use well established tables of planetary and lunar motion.

[jjr495]

The defense's reply states in the first footnote that they have received material from Owen. I sure hope it is better than the tech specs on his EVB software site. It must have been written by a non-native English speaker. It is largely unintelligible.
After looking at Owen's CV I conclude that it is very unlikely that he wrote or understands the algorithms that his software uses. The techniques mentioned in the tech specs of Owen's software are standard techniques, and well-tested for speaker id. Under many conditions that probably would not be considered new science. However, there are stacks of papers that report the serious degradation of these techniques in distressed speaker, poor recording conditions, and with limited speaking times. These are areas of active research to improve speaker id systems. There is probably no research that has been done on speaker id of speech that has degraded to the point of being unintelligible (with the exception Reich of course).  Also, as I said before, how does Owen even know that he is not mixing different voices when he does his analysis.

[cboldt]

I was searching for something else, and ran across West's December 5, 2012 Motion to Compel Regarding Voice Identification.

I mention this here to put the importance (or lack of importance) of the expert opinions.  I had forgotten that neighborhood eyewitnesses, other than John, had given statements that they recognize the voice screaming for help, as heard on the recorded 911 call, as that of Zimmerman.

[nomatter_nevermind]

I had forgotten that neighborhood eyewitnesses, other than John, had given statements that they recognize the voice screaming for help, as heard on the recorded 911 call, as that of Zimmerman.

The witnesses are W-45 and his unidentified neighbor, according to W-45's FBI interview summary.

W-45 is 110% sure. The neighbor is only 100% sure.

ETA: There might be others, but West said 'two neighbors' in the motion, and referenced a single FBI report.

ETA 2: W-45 apparently doesn't socialize with Zimmerman, but he described positive interactions with him as a neighbor. He's not an entirely unbiased witness.

[RickyJim]

ETA 2: W-45 apparently doesn't socialize with Zimmerman, but he described positive interactions with him as a neighbor. He's not an entirely unbiased witness.

I've heard how sure some of these people are.  (I think Boobie said he would swear on a stack of bibles that it was Trayvon.)  But I haven't heard of a single one that claims he heard the subject scream like that before or who picked the screams out of a lineup.  And we demand a Frye hearing for scientists but nothing for these kind of witnesses?  The fact that West wrote a memo shows how much depends on calculations involving how dumb a jury they will get versus getting good evidence.   

[cboldt]

I've heard how sure some of these people are.  (I think Boobie said he would swear on a stack of bibles that it was Trayvon.)  But I haven't heard of a single one that claims he heard the subject scream like that before or who picked the screams out of a lineup.  And we demand a Frye hearing for scientists but nothing for these kind of witnesses?  The fact that West wrote a memo shows how much depends on calculations involving how dumb a jury they will get versus getting good evidence.

IIRC, Robert Zimmerman, Sr. said, under oath, that he'd heard his son George scream for help before.

"It was absolutely George's" voice screaming on the tape, his father testified, saying he had heard his son yell "many times" before.

Clashing views of Zimmerman arise in bail hearing - Miami Herald - 29 June 2012

The experts come to the case without any familiarity with the various speakers, beyond what they are given for evidence to work with.  The Frye test just flat out doesn't apply to a fact witness, because the fact witness isn't depending on the application of science, technology, or specialized expertise.  That doesn't mean a fact witness can't be challenged as having insufficient basis for the statement.  Cutcher, for example, didn't see the altercation, and doesn't know the natural voices of either Martin or Zimmerman, but she claims Martin was the screamer because the screams sounded like a little boy, and she saw Zimmerman standing over Martin some half a minute or more after the shot was fired.

West's motion was well taken, because it aims to use the power of the court to prospectively force the state to disclose specific exculpatory evidence.  It also informs the state that the defense is interested in the conclusions of at least one other fact witness who may be qualified by familiarity with Martin's voice - Witness 8.

[RickyJim]

Robert Zimmerman or anybody else who claims to be able to identify the screamer from the recording would have to be closely questioned on the previous instances they heard him scream like that and how many other people they had heard scream in that manner so they could tell the difference.  When GZ himself tried to recreate the screams, he didn't come close.  The pitch was much lower.  The only thing that can be determined with certainty from the 911 call is the extreme distress the yeller was in which indicates he had a serious concern for his life.  That it can't be disproved that Zimmerman was the screamer, plus the reasonableness of the concern (EG, witness#6 last saw the wrestlers on the concrete with GZ on his back) can only mean an acquittal.

[cboldt]

When GZ himself tried to recreate the screams, he didn't come close.  The pitch was much lower.

I didn't find the pitch to be much different, but the duration sure was.  If SPD was serious about trying to get a match that would facilitate scientific analysis, they'd play the recording for Zimmerman as a cue.

Anyway, my general point was only to explain that the Frye test goes to the science or expertise that is played against the evidence.  This is different from what a fact witness is allowed to base testimony on.  I don't know of a simple test for the reliability of fact witnesses, other than the hearsay rule.

[RickyJim]

Hmm.  AJ could tell us how close he got.  IIRC I thought he was an octave away from some of the 911 screams.  Of course O'Mara could ask an expert to see if could get George to match the screams.  Then the prosecution would demand a Frye hearing to see if that was admissible.     

[AJ]

Hmm.  AJ could tell us how close he got.  IIRC I thought he was an octave away from some of the 911 screams.  Of course O'Mara could ask an expert to see if could get George to match the screams.  Then the prosecution would demand a Frye hearing to see if that was admissible. 

I don't have the example screams Mr. Zimmerman did, and taking a youtube video as a source wouldn't be a wise idea (no clue how many times it has been re-encoded, etc). I have the video that gzlegalcase released (you now have to do a media request to get it because it's 600MB), but it doesn't include the screams. Also, on something like that I probably wouldn't give an estimation of how close they are; I'd probably just submit screenshots of the oscilloscopes/spectrographs of each and let each person come to their own conclusion.

[nomatter_nevermind]

I think Boobie said he would swear on a stack of bibles that it was Trayvon.

9/284

[AghastInFL]

The witnesses are W-45 and his unidentified neighbor, according to W-45's FBI interview summary.

W-45 is 110% sure. The neighbor is only 100% sure.

ETA: There might be others, but West said 'two neighbors' in the motion, and referenced a single FBI report.

ETA 2: W-45 apparently doesn't socialize with Zimmerman, but he described positive interactions with him as a neighbor. He's not an entirely unbiased witness.

W-11also identified George by name as the one screaming - Help as he was being pummeled.