Speech-based lie detection in Russia
« previous post | next post »
Andrew E. Kramer, "Russian A.T.M. With an Ear for the Truth", NYT 6/8/2011:
Russia’s biggest retail bank is testing a machine that the old K.G.B. might have loved, an A.T.M. with a built-in lie detector intended to prevent consumer credit fraud.
Consumers with no previous relationship with the bank could talk to the machine to apply for a credit card, with no human intervention required on the bank’s end.
The machine scans a passport, records fingerprints and takes a three-dimensional scan for facial recognition. And it uses voice-analysis software to help assess whether the person is truthfully answering questions that include “Are you employed?” and “At this moment, do you have any other outstanding loans?”
The voice-analysis system was developed by the Speech Technology Center, a company whose other big clients include the Federal Security Service — the Russian domestic intelligence agency descended from the Soviet K.G.B.
Dmitri V. Dyrmovsky, director of the center’s Moscow offices, said the new system was designed in part by sampling Russian law enforcement databases of recorded voices of people found to be lying during police interrogations.
I wrote about an earlier NYT story of the same sort almost seven years ago — "Analyzing voice stress", 7/2/2004:
Yesterday's NYT had an article on voice stress analyzers. As a phonetician — someone who studies the physics and physiology of speech — I've been amazed by this work for almost three decades. What amazes me is that research (of a sort) and commerce (at a low level) and law-enforcement applications (here and there) keep on keepin' on, decade after decade, in the absence of any algorithmically well defined, reproducible effect that an ordinary working speech researcher like me can go to the lab, implement and test.
Well, these days there's no need to go to the lab for this stuff — you just write and run some programs on your laptop. But that makes the whole thing all the more amazing, because after 50 years, it's still not clear what those programs should do. I'm not complaining that it's unclear whether the methods work — that's true too, but the real scandal is that it's still unclear what the methods are supposed to be.
Specifically, the laryngeal microtremors that these techniques depend on haven't ever been shown clearly to exist, as far as I know. No one has ever shown that if these microtremors exist, it's possible to measure them in the pitch of the voice, in a way that separates them from all the other phenomena that modulate the pitch at similar rates. And that's before we get to the question of how such undefined measurements might be related to truth-telling. Or not.
How can I make you see how amazing this is? Suppose that in 1957 some physiologist had hypothesized that cancer cells have different membrane potentials from normal cells — well, not different potentials, exactly, but a sort of a different mix of modulation frequencies in the variation of electrical potentials between the inside of the cell and the outside. And further suppose that some engineer cooked up a proprietary circuit to measure and display these alleged variations in "cellular stress" (to the eyes of a trained cellular stress expert, of course), and thereby to diagnose cancer, and started selling such devices to hospitals, and selling training courses in how to use them. And suppose that now, almost half a century later, there is still no documented, well-defined procedure for ordinary biomedical researchers to use to measure and quantify these alleged cell-membrane "tremors" — but companies are still making and selling devices using proprietary methods for diagnosing cancer by detecting "cellular stress" — computer systems now, of course — while well-intentioned hospital administrators and doctors are occasionally organizing little tests of the effectiveness of these devices. These tests sometimes work and sometimes don't, partly because the cellular stress displays need to be interpreted by trained experts, who are typically participating in a diagnostic team or at least given access to lots of other information about the patients being diagnosed.
This couldn't happen. If someone tried to sell cancer-detection devices on this basis, they'd get put in jail.
But as far as I can tell, this is essentially where we are with "voice stress analysis."
And seven years later, things haven't really changed. Of course, not all such systems claim to work on the basis of laryngeal micro-tremors — some explicitly say proudly that they *don't* use that method, which has gotten a bit of a reputation as snake oil. But neither do they cite a different method that's well enough specified for someone else to implement it and test it.
Today's NYT article describes the new Russian system in typically vague terms:
But the Speech Technology Center says even people who know about the voice-stress program would have trouble fooling it, because it measures involuntary nervous responses, the way a polygraph test does.
The center’s director, Mr. Dyrmovsky, said the voice-stress system analyzed vibrations as shaped by the contours of an individual’s throat, larynx and other tissue involved in speech. When a person becomes agitated, he said, involuntary nervous reactions alter these shapes, changing the tone and pacing of speech.
This is basically double-talk: to say that the system "analyze[s] vibration as shaped by the contours of an individual's throat, larynx and other tissues involved in speech" is just to say that the system analyzes speech. Of course, this quote is filtered through the reporter, Andrew Kramer, so the vacuity of the explanation is not necessarily the fault of Mr. Dyrmovsky.
This "voice-stress program" is not one of the products listed on the Speech Technology Center's English website, nor have I found any other clues to how their voice-based lie-detector actually works. Pending future revelations, I'm going to assume that the present is like the past, and that this is more of the same old same-old.
Some previous LL posts on speech-based lie detection:
"Analyzing voice stress", 7/2/2004
"Determining whether a lottery ticket will win, 99.999992849% of the time", 8/29/2004.
"KishKish BangBang", 1/17/2007
"Industrial bullshitters censor linguists", 4/30/2009 (see especially the comments threads, e.g. here, here, here, here.)
What I really want to know about the Speech Technology Center is whether any of its personnel did time doing speech research in the Marfino sharashka during the early 1950s, along with Aleksandr Solzhenitsyn, who fictionalized his experiences as a sort of morality play in The First Circle, and Lev Kopelev, the original of Solzhenitsyn's Rubin, who wrote about it in his memoir Ease My Sorrows. Or, more likely, whether any of them were trained by Marfino veterans.
Jethro said,
June 8, 2011 @ 10:33 pm
"[i]t measures involuntary nervous responses, the way a polygraph test does."
This is probably more true than the reporter realizes.
[(myl) I wish. The physiology of polygraph inputs — blood pressure, skin conductance, etc. — is pretty well understood, and there are well-defined and reasonably reproducible techniques for measuring them. This is not even close to true for the speech-based systems, where there have never been any clear definitions of reproducible inputs (other than the speech waveform itself). Instead we're told about proprietary methods for measuring vaguely-described things like "micro-tremors", for which there has never been a credible published algorithm (that I'm aware of).
The interpretation of polygraph data is more controversial, but at least it's clear what people are worrying about the interpretation of. That's not even true for the speech-based measures (again, as far as I've been able to determine in 40 years of fairly serious looking).]
Dakota said,
June 9, 2011 @ 2:07 am
This probably works in the same way a placebo works — only if you believe in it (and placebos have been documented to work).
I am reminded of the old lo-tech Norse custom of determining truth-telling by lifting a piece of sod in the middle while leaving it attached to the earth at both ends (it must have been a fairly large piece) and having the disputing parties pass under it. It was presumed the lying party would have something bad done to them by the pagan gods they had just flaunted by profaning the earth. An instance of this is recorded in the sagas. When it was the turn of the lying party to pass under the sod, they tried to game the system by "accidentally" knocking down the supporting pole. Result? Truth was determined as accurately as when King Solomon threatened to cut a baby in half in order to determine its true mother.
Henning Makholm said,
June 9, 2011 @ 6:55 am
The cancer-detection analogy is remarkably detailed. Was it supposed to remind us of anything? (Apart from acoustic lie detection, that is).
[(myl) No, it was completely fabricated. I hope it was, anyhow.]
Mr Punch said,
June 9, 2011 @ 11:04 am
Dakota's right. The Russians once touted a machine that supposedly eliminated people's need for sleep, but the device turned out to work just as well when it wasn't plugged in.
army1987 said,
June 9, 2011 @ 12:27 pm
@Dakota: In Rome there was an even lower-tech way: a statue where you was supposed to put your hand and if you lied it would magically be cut off. People were just too scared to do that.
Keith M Ellis said,
June 9, 2011 @ 3:49 pm
Mark, your point that at least with polygraph testing there is actually something everyone agrees is data that might be, er, interpretable is a good one.
Still, the unfortunate truth that even with that advantage polygraph testing and its relationship to lie detection is a pseudoscience that, nevertheless, is widely accepted in the US as reliable, even by law enforcement in some cases, is revealing of just how much deeply vested interests many have for believing in and promoting lie-detection technology. I am sad to opine that this is not the kind of enterprise that is likely to ever be much vulnerable to hostile critiques from credible scientists and such.
Benjamin said,
June 9, 2011 @ 6:55 pm
Actually, Mark wasn't too far off on the cancer thing.
http://en.wikipedia.org/wiki/Royal_Rife
And you would be amazed at the pseudoscience that goes unchecked by the law. Even Robert Young (http://en.wikipedia.org/wiki/Robert_Young_%28author%29) has managed to keep on with little problem.
[(myl) Well, at least he's been taken to court a few times, e.g. (from the Wikipedia article):
And he's selling books and advice, not devices. If he were selling devices — prick your finger and put a blood sample into the Young-o-meter attached to a PC running the sekrit Young-o-matic analysis software that diagnoses early-stage cancer to be eliminated by dietary changes — the FDA (I trust) would come down on him much harder.]
Helena Constantine said,
June 9, 2011 @ 7:05 pm
" No, it was completely fabricated. I hope it was, anyhow"
Is this close enough?
http://www.sciencebasedmedicine.org/index.php/dr-christiane-northrup-and-breast-thermography-the-opportunistic-promotion-of-quackery/#more-7454
Benjamin said,
June 9, 2011 @ 7:13 pm
@Helena Constantine
You know, I keep thinking Gorski's done a post on Rife, but I can't find anything devoted to him on SBM.
tpr said,
June 10, 2011 @ 8:49 am
A device manufactured in the UK for use detecting bombs at military checkpoints in Iraq was recently shown to be nothing more than a divining rod with an electronic theft-prevention tag attached to it. The devices which sold for $US 40,000 apiece were completely useless and may have contributed to the deaths of hundreds of people. The manufacturer has now been arrested on fraud charges.
http://news.bbc.co.uk/2/hi/uk_news/8476381.stm
Dakota said,
June 12, 2011 @ 7:04 am
@army1987 : There is also the mom jabbar, "the ability of a mother to discern the truth in any situation", probably as intimidating as its namesake the gom jabber, a fictional box containing a poison needle the Bene Gesserite used to ascertain a person's humanity by having them stick their hand in it.
http://encycl.opentopia.com/term/Gom_Jabbar
Some of these plots seem to be retreads.
Ray Girvan said,
June 13, 2011 @ 5:52 pm
> Gom jabbar
I suppose everyone's seen this?