Fla. judge to decide on whether 911 scream analysis is admissible in Zimmerman trial

Chuck Ross Investigative Reporter
Font Size:

Whether or not voice identification experts will be allowed to testify in the George Zimmerman trial will be determined in a Florida courtroom today.

Circuit Judge Debra Nelson will hear arguments for the admissibility of testimony from experts listed on Florida prosecutors’ list of potential witnesses.  These experts claim to be able to offer analysis of a scream heard in the background of a 911 phone call that also captured the sound of the gunshot that killed Trayvon Martin in February 2012.

In court, Zimmerman’s defense team will try to argue that the methods used by the state’s potential witnesses are not proven and reliable under Florida’s Frye standard for evidentiary admissibility.

Zimmerman attorney Mark O’Mara wrote to the court that under Frye, “the ‘community acceptance’ standard exists to act as a valve to safeguard against the presentation of evidence by scientific methods that are not provably sound.”

If identified, that scream is pivotal to the case, set to begin June 10.

But academics in the field of voice identification tell The Daily Caller that the science behind matching voices to screams is still unproven.

Sadaoki Furui of the Tokyo Institute of Technology — and winner of Japan’s Royal Order of the Purple Ribbon award for his work in the field voice identification technology — told The Daily Caller that “there is no scientific or engineering method to correctly match a voice to a known scream.”

The scream voice is susceptible to psychological conditions and background noise, he said, “and it is almost impossible to make a good statistical model to cover such variation in each person.”

John Hansen heads the Center for Robust Speech Systems at the University of Texas – Dallas.  He agreed with Furui’s assessment and added that for voice identification analysis to be accurate “there needs to be some evidence based research that explores the variability of speaker identity across a range of screams from the same subject.”

He added, “the technology which has evolved successfully for speaker ID / verification has not been evaluated for scream detection in a scientific manner.”

Zimmerman’s defense has maintained the same argument.

If Zimmerman was the screamer, his defense that he was being viciously attacked by Martin would be supported.

Zimmerman has maintained that when he shot Martin at nearly point-blank range in the chest, Martin was on top of him, hitting him in the face and bashing his head against the ground “MMA style.” Zimmerman told police that he was screaming for help during the attack but that nobody would come to his aid.

Martin screaming for help would suggest that Zimmerman shot the teen in cold blood after relentlessly pursuing the “suspicious”-looking teen, even after he was told to stand down by a non-emergency dispatcher.

The second-degree murder charges pending against Zimmerman require that he was of a “depraved mind regardless of human life” and will be easier to prove to the jury if he is seen as having shot a victim pleading for his life.

Last year the Orlando Sentinel reported the opinions of two men who work in the field of voice forensics and voice identification.  Using two different techniques, Tom Owen and Ed Primeau determined that the crucial screams for help were not elicited by Zimmerman.  Both men are listed on Florida prosecutors’ expert witness list.

Owen told the Sentinel that he used his proprietary software, Easy Voice Biometrics, to determine that Zimmerman’s scream was not the one heard on the 911 recording. Owen said that he would expect a 90% match given the quality of the recorded scream but found that it provided only a 48% match with a known Zimmerman voice exemplar.

This, he told the Sentinel, allowed him to dismiss Zimmerman as the screamer.  Owen has pointed to a case in Connecticut in which his analysis using his new software helped lead to the conviction of Sheila Davalloo.  In that case, Owen matched Davalloo’s voice to a 911 phone call, though that call did not contain a scream or yell.

Primeau used a naïve analysis — his own ear — to determine that the scream came from Martin.  “That’s a young man screaming,” he told the Sentinel.

The Sentinel as well as The Washington Post contacted other analysts in the field of voice identification to match the voice of the screamer.

Alan Reich is a New Jersey based acoustic forensics expert contacted by The Post who believes that Martin was screaming for help while Zimmerman’s “side-bars” sounded like an “evangelical preacher” or “carnival barker.”

Also contacted were James Harnsberger and Harry Hollien of the University of Florida.  Both were unable to make definitive determinations about whose voices were heard in the call.

Reached for comment, Zimmerman co-counsel Don West told The Daily Caller that the different methodologies and conclusions of voice identification analysts will only confuse the jury.  “This could contribute to an unjust verdict if the jury were to give this evidence more importance than it deserves,” he said.

Besides conflicting reports by different experts, West says that the experts’ findings “are inconsistent with the physical evidence and the testimony of an eyewitness who heard George Zimmerman crying out for help while being beaten by Trayvon Martin.”  Photographs taken at the crime scene do show a bloodied and swollen Zimmerman.

West also pointed out that FBI analysts were unable to determine who was heard screaming.  The Post article last year citing Reich’s findings also reported that former FBI analyst James Ryan considered the quality of the 911 recording to be too degraded to make a reliable determination about the voices in the background.

West wrote to the court that though voice identification methods have been used in criminal court cases in the past, “the defense knows of no cases where speaker identification testimony was admitted involving the comparison of speech exemplars to short bursts of yelling or screaming.”

James Wayman a voice biometrics researcher at San Jose State University told The Daily Caller last year about Owen’s and Primeau’s findings, “I would be willing to testify against the admissibility of any of this evidence as ‘scientific’ under the Daubert criteria as required in federal courts.”

The Florida legislature recently voted to replace its Frye standard with Daubert, bringing the number of Frye states down to nine.  But the change won’t occur until July 1.

Generally, the Frye standard, which Zimmerman’s attorneys are attempting to invoke, provides a lower threshold for expert testimony and opinion than Daubert which became the federal standard after a 1993 Supreme Court ruling.  It is also used in many state courts.

Christopher Brown, a Florida defense attorney says “the way courts have interpreted Frye in Florida, they have opened the door to allow pretty much anyone who is nominally an expert in the area at issue to testify.”

Diane Mazur, professor of law at the University of Florida said, “It’s unlikely these experts can show their methods are reliable when tested under ‘blind’ circumstances, and that should be the standard for admission.”

Mazur believes that the expert testimony should be excluded from the trial but that people familiar with either Zimmerman or Martin can identify the voices at trial.  “There’s no reason why witnesses or jurors can’t compare voices themselves,” she said.

Brian J. Foley, professor at Florida Coastal School of Law said, “the bottom line is that the more unusual the scientific evidence, the less peer testing and the like that it has been subjected to, the less likely a Florida judge will let a jury hear it.”

Reached for comment, Bill Eddins, president of the Florida Prosecuting Attorneys Association said that it is inappropriate for his organization to comment on a pending court case.