JEFFERSON CITY, Mo. – Rumors concerning the extramarital affair and alleged blackmailing of a married woman by Gov. Eric Greitens continue to spread across the state and nation like wildfire less than 48 hours after the report was issued by KMOV. And the key piece of evidence that seems to matter in the conversation is a recorded conversation between the woman and her now-ex-husband.
Some are questioning the authenticity of the tape, asking whether it really is the voice of the woman who has been identified in the claims.
Neither the Governor nor his criminal lawyer has said that was the woman on the tape. Her husband said it was her,” host Scott Faughn said on the latest taping of This Week in Missouri Politics. “ If that tape is verified, I guess the Governor isn’t saying she’s lying.”
“That and that’s sort of where this lies. No pun intended,” Jane Dueker responded. “Either the tape was made up… the other option, it was her on the tape and she was lying. Only she can say whether what she said to her husband is true or it’s not true. Those are facts.”
But how can we truly know?
In court, the authenticity of an audio recording may be called into question, meaning that it would be up to the forensic audio experts to determine whether the evidence has been altered by confirming the integrity of the recording.
The job of a forensic audio expert usually consists of three parts: enhancing the tape to make the voices more easily heard, looking to see if the tapes have been tampered with, and identifying the voices in the recording.
The first part has no effect in proving authenticity. The others, however, play every part.
To catch tampering in an audio file, audio experts would look for signs of alteration, like abrupt changes in the ambiance, examining the file to look for the constant hum that is created by the electrical grid. They also evaluate the volume and tone of the voice, looking for technical details like an unnatural waveform in the audio.
But the other part of the process may be the one that is more important to those doubting the validity of the recording.
“If the tape is wrong – and there’s two people we know of that know this woman intimately – her husband and the Governor of the State of Missouri,” Faughn said. “Her husband says it is. The Governor’s criminal defense lawyer has said it’s not. If it’s true, there are ways you can verify the tape.”
“There are ways,” Rep. Kathie Conway said. “There’s voice recognition.”
Voice identification does indeed to be the most likely answer to proving whether the recording is truly of the woman in question, but the first component needed for that is an example to compare the file against.
As the woman has not come forward, and her identity has not been revealed, at least not to the public, that would be a nearly impossible feat to manage. In theory, if someone has knowledge of the woman in question, then all one would need is to record her speaking and provide that recording, along with the original recording, to a forensic expert and verify whether the voices match.
If, however, a file were readily available, all that would need to occur is for the examiners to take the recording in question, as well as a recording of the known person’s voice, and compare the two files using three tests: a spectrograph analysis, an average pitch analysis, and a statistical analysis, which involves a database of millions of voices.
One the two samples are run through the program, examiners are given a percentage from 0 to 100 evaluating the likelihood that the voices are the same. Examiners also might use their judgment to compare the accents, syntax and breathing patterns of the person. Using such methods, the expert would listen for the speaker’s choice of words, structure sentence, usage of “um” or “like” or even speech difficulties – which when all combined together is called the idiolect – a person’s specific and individual way of talking.
An example of how this all might be used is to look back at the Trayvon Martin case in 2012 when forensic experts analyzed the recording of a phone call to police by George Zimmerman. They attempted to determine whether a scream in the background came from Zimmerman or Martin. Zimmerman told police he had fired in self-defense. In an analysis by the Orlando Sentinel, the scream was a just 48 percent match to Zimmerman, and the analyst determined it could not have been made by Zimmerman.
Another example of how voice recognition is used would be when experts from around the world tried to identify the masked terrorist known as Jihadi John who was shown in the video-captured beheading of journalist James Foley.
Even with all of this technology, that doesn’t mean that courts take these analyses as definitive proof. They’re often open to challenges on both the grounds of the methods used and reliability. In truth, the scientific community has explicitly discredited some voice analysis techniques and is still far from reaching a consensus on what the most effective method for identifying voices is.
The Scientific Working Group on Digital Evidence, led by representatives from the U.S. Secret Service and the FBI, provides peer-reviewed guidelines for audio and video forensics engineers, and the Secret Service is calling for standards that all audio forensics labs must meet.
The quickest way, however, stills remains this: the woman could come forward and confirm that the voice on the recording is, in fact, hers.
That finally came about Friday afternoon, when the law firm representing the woman, Knight & Simpson, issued a press release, which stated that “it is very disappointing that her ex-husband betrayed her confidence by secretly, and without her knowledge, recording a private and deeply personal conversation and then subsequently released the recording to the media without her consent.”
Just got a statement from a law firm representing the woman who accused Greitens of blackmail in a recording made by her ex-husband. She has declined to give interviews. pic.twitter.com/cwvRP4IuGY
— Matt Pearce 🦅 (@mattdpearce) January 12, 2018
She did not recant or challenge anything that was said in the recording.