A peer-reviewed journal has recently yanked a published paper from its Web site after the makers of a voice-analysis system--which is sold as a device to detect emotional stress and help ferret out liars--complained the article contained inaccuracies and defamed them.
Penned by phoneticians Anders Eriksson of the University of Gothenburg and Francisco Lacerda of Stockholm University, both in Sweden, the controversial paper appeared in the December 2007 issue of The International Journal of Speech, Language and the Law with the title "Charlatanry in forensic speech science: A problem to be taken seriously." The paper argues that there is no scientific basis for techniques that aim to determine emotional stress by analyzing the sound of a voice. In particular, Eriksson and Lacerda take aim at the Layered Voice Analysis (LVA) systems made by Nemesysco Ltd., of Netanya, Israel, of which they write, "The ideas on which the products are based are simply complete nonsense."
Amir Liberman, the founder and CEO of Nemesysco, took exception to the paper. On 3 November, his lawyers wrote to the journal's publisher, Equinox Publishing Ltd. of London, threatening to sue for defamation if the article were not retracted. Liberman says that he acted not because the researchers questioned his technology but because they targeted him personally by, for example, noting in a section titled "Who is Mr. Liberman?" that he has no university degree. "The objection was not in the publication of their study results, it was in their calling us charlatans," Liberman says.
On 4 December, the journal removed the article from its subscription-based Web site and posted a notice that acknowledged that "Mr. Liberman and Nemesysco Limited ... were not invited to comment on the article prior to its publication where, in view of the content of the article, it would have been appropriate to invite them to do so." The journal's action has Lacerda crying foul. In a letter to Science, he says that "Now Nemesysco's resources have grown so large that they could even force the withdrawal of a peer-reviewed paper questioning its technology."
Janet Joyce, managing director at Equinox, declined to discuss the specifics of the case, but she says the journal--which is published biannually, has a circulation of less than 500, and employs no full-time staff--simply lacks the resources to put up a legal fight. The journal has agreed to publish a rebuttal letter from Lieberman and the company, but Joyce notes that "we didn't withdraw the article. It's still in print."
In the paper, Eriksson and Lacerda cite two studies that question the scientific validity of LVA technology, which analyzes the pressure waves that make up an utterance and tallies spikes, or "thorns," and plateaus in order to deduce the speakers state of mind. The phoneticians also tried to reproduce the software for the LVA system using computer code in Liberman's 2003 patent as a guide. Lacerda claims that the spikes and plateaus are largely artifacts of the digitization of the sound and that the system makes arbitrary connections between the rates at which they occur and emotional states such as "high stress" or "untruthfulness." "The measurements themselves ... don't contain any meaningful information," Lacerda asserts.
Liberman counters that Eriksson and Lacerda used information from only one of three patents and that they never used one of Nemesysco's systems itself. "This attack is being made by people who never saw our technology, never touched the equipment," he says.
The debate notwithstanding, Nemesysco's system is already in use. In an effort to stamp out fraud, 25 local government councils in the United Kingdom are using it in conjunction with scripted questions to evaluate phone callers applying for housing and other benefits. (Applicants can potentially increase their benefits by claiming to be poorer than they actually are.) The national government will decide in the spring of 2010 whether to deploy the technology nationwide, says John Stevenson, a spokesperson for the Department of Work and Pensions. Officials use the system to decide whether further investigation is warranted, he says: "They wouldn't just say from one phone call we're stopping the benefit."
A longer version of this story will appear in the 13 February issue of Science.