So, for example, one of these so-called experts used a technique called "biometric voice analysis" to compare Zimmerman's voice, recorded when he was calmly speaking to police dispatchers shortly before the shooting, and the screaming voice on the 911 tape. This expert found a 48 percent match between the voices, concluding "with reasonable scientific certainty that it's not Zimmerman." First off, what is biometric voice analysis? Is it similar or different from voice identification through voice spectrography, commonly called a "voiceprint," which has been studied for years but which most courts have not admitted into evidence because it lacks scientific reliability and acceptability. Another so-called expert gave the media his "strong opinion" that the voice was Trayvon Martin's "without a doubt." What was the basis for this expert's opinion? "The tone of the voice is a giveaway," he said. "That's a young man's voice." This from an expert?
Some of the opinions by voice experts raise similar concerns. For example, whereas fingerprint and DNA evidence have unique and distinguishable characteristics, there does not appear to be any scientific basis for claiming, as one of the above voice experts did, that each individual's voice is unique. And anyway, there are so many marked differences that may occur among various groups of talkers that giving an opinion "to a reasonable degree of scientific certainty" is blatantly misleading. Even using terms like "probable" or "inconclusive" would be problematic, but a much more accurate opinion. Also, comparing a speaker's voice in a calm state and an emotional state, as one of the voice experts did, makes any informed comparison even more hazardous. Finally, published studies of voice identification under experimental conditions show very high rates of error. Indeed, a report by the National Academy of Sciences on forensic voice identification concluded that the scientific basis for making reliable voice identifications is weak. The FBI, as a result of the report, apparently does not use forensic voice analysis for courtroom evidence, although as with polygraph evidence, it continues to use it for investigative purposes.
If George Zimmerman is brought to trial, the voice of the person screaming probably will be a major issue. There likely will be contradictory proof of whose voice it is. Such proof can readily be admitted from persons who are familiar with that person's voice. But such proof should not be admitted based on the opinion of a forensics expert. It's just not reliable.