In deciding whether to admit the voice-recognition technology used by prosecution audio experts Tom Owen and Alan Reich, Nelson had to determine whether it is too novel or whether it has been accepted by the scientific community at-large.
"There is no evidence to establish that their scientific techniques have been tested and found reliable," the judge said in her ruling.
Owen was hired by the Orlando Sentinel last year to compare a voice sample of Zimmerman with screams for help captured on 911 calls made by neighbors. He said Zimmerman's voice doesn't match the screams. He only compared Zimmerman's voice to the 911 calls because he didn't have a voice sample for Martin at the time.
"The screams don't match at all," Owen testified during the hearing. "That's what tells me the screams aren't George Zimmerman."
Owen also testified that remarks Zimmerman made in a conversation with a police dispatcher aren't a racial slur. He testified Zimmerman said, "These f------ punks."
Reich testified in a report for prosecutors that the screams on the 911 tapes were from Martin, and the defense does not want him to testify at trial.
Reich's analysis also picked up words that other experts couldn't find. They include the words, "This shall be" from Zimmerman and "I'm begging you" from Martin.
Reich's testimony would "confuse issues, mislead the jury," the judge said.
In contrast, a British audio expert testified for the defense that it would be extremely difficult to analyze voices by comparing screaming to a normal voice.
"I've never come across a case in my 13 years where anybody's tried to compare screaming to a normal voice," said audio expert Peter French.
A second audio expert for the defense, George Doddington, also criticized prosecution experts who said Friday that screams and pleas on a 911 recording likely belonged to Martin.
"It's all ridiculous," Doddington said.
Follow Kyle Hightower on Twitter at http://www.twitter.com/khightower .
Follow Mike Schneider on Twitter at http://twitter.com/MikeSchneiderAP .