By Barbara Liston
SANFORD, Florida (Reuters) - With only two days left before trial, lawyers for accused murderer George Zimmerman are trying to block testimony suggesting his victim was the person screaming for help in the background of a 9-1-1 recorded call, moments before Trayvon Martin, an unarmed 17-year-old, was shot to death in 2012.
A lengthy hearing which began Thursday for defense lawyers to challenge the methods used by prosecution audio experts will be carried over for a special Saturday session in a central Florida court.
Zimmerman is set to go on trial Monday on a charge of second degree murder. Lead defense attorney Mark O'Mara has called the 9-1-1 recording "the most significant piece of evidence in the case."
In 2012, Zimmerman was a neighborhood watch captain in a gated community in Sanford, near Orlando, when he shot Martin, a black teenager and guest of a resident. Prosecutors say Zimmerman profiled Martin as a potential criminal, then pursued and confronted him.
If the identity of the person crying for help can be established, that would indicate who was dominant in the deadly fight, an important issue to Zimmerman's claim he shot in self defense.
At issue was whether the expert analyses were derived from techniques generally accepted in the scientific community, or some new, untested approach.
Two prosecution audio experts in testimony Thursday and Friday qualified their findings, saying the quality of the recording was poor. One acoustics consultant, Alan Reich tentatively identified Martin's voice, including what he said was a final "stop" yelled by Martin just before the gunshot.
"The words at scream level were almost entirely Trayvon Martin," Reich testified on Friday.
Another forensic audio expert, Thomas Owens, tentatively ruled out Zimmerman based on a comparison of the screams to known samples of Zimmerman's voice.
"The screams don't match at all. That is what tells me it's not George Zimmerman," Owen testified Friday.
Experts from Forensic Communications Associates, who did not testify, reported cautiously that both Martin's and Zimmerman's voices were recorded during the struggle.
A fourth expert, FBI analyst Hirotake Nakasone, reported he could isolate only 2.3 seconds of unobstructed screaming, not enough to attempt an opinion. Nakasone testified on Thursday that the FBI standard for making a positive identification is a recording of at least 16 words and 20 seconds.
Defense lawyer Don West indicated the defense has two of its own experts, however, their findings have not been made public.
Nakasone testified voice identification is an inexact science, and that the sound of someone's voice is effected by emotion, circumstances and even air quality. All of the experts said it was not unusual for audio experts to reach different conclusions in their analyses.
All three described using traditional techniques such as critical listening and computer biometric software in their analyses. Owens testified his software is relatively new but based on long used algorithms.
West said he will present three witnesses on Saturday.
(Editing by David Adams, Bernard Orr)