November 30, 2002	Saturday	Ramazan 24, 1423

PARIS: On the basis of scientific tests, a French TV channel has established that the voice on a Nov 12 audiotape broadcast by Al-Jazeera is not that of Osama bin Laden, and this in spite of a recent CIA “confirmation” that it belongs to the Al Qaeda leader.

The revelation was made on Thursday in its news report of France-2, the country’s most-watched public TV channel, anchored by David Pujadas.

The voice that was broadcast on Nov 12 had the Al Qaeda leader claiming that terrorist attacks would be undertaken against those countries — notably France, but also Great Britain, Germany and Israel — that had chosen to support the United States in its campaign against terrorism in general and Al Qaeda in particular.

In order to determine whether the voice might not be that of Osama bin Laden, France-2 went to Lausanne, Switzerland, to one of the world’s leading specialists in speech recognition, Professor Herve Bourlard, who is with the Dalle Molle Institute for Perceptual Artificial Intelligence (IDIAP), as well as a director of the Swiss Federal Institute of Technology in Lausanne.

Professor Bourlard was given some two hours’ worth of tape — much of it broadcast on Al-Jazeera — said to contain Osama’s voice, including the Nov 12 sample which the CIA said recently was, according to its own tests, definitely that of the Al Qaeda chief.

After several tests which involved establishing the “fineprint,” of the voices, that is, breaking them down into digital data, Professor Bourlard was able to establish that although most of the samples were indeed of Osama’s voice, that the Nov 12 tape as broadcast on Al-Jazeera was of a completely different voice.

Professor Borlard was able to produce a graphic representation of the voices, which indicated that the Nov 12 Al-Jazeera sample was of an entirely different voice, indeed of a voice that might have been artificially generated to leave the impression it was that of Osama bin Laden.

Queried as to whether the mediocre quality of the Nov 12 sample might be responsible for his results, he noted that the origins of the samples — which included Osama’s voice on a walkie-talkie and portable telephone — had nothing to do with his conclusion, especially as some of the samples that had been proved to be those of the Al Qaeda leader were of even more mediocre quality.

Given the existence of similar technology at other speech recognition laboratories around the world, Professor Bourlard said he was surprised that no other specialist had been called upon — outside of those at the CIA — to analyze the voice samples to see whether the Nov 12 representation of Osama’s voice was authentic.

Professor Bourlard noted that statistically speaking, his method had a margin of five per cent error, that is, there was one chance in 20 that his own conclusions might be wrong.

The France-2 crew that undertook the investigation noted that all of the French laboratories they contacted had refused to take part in their investigation, with a suggestion that it was not in their best interests to “go against the grain” and attempt to second-guess the results of US governmental tests that claimed to firmly establish that the Nov 12 Al-Jazeera voice was indeed that of Osama.

Professor Bourlard is an internationally-known specialist in the field of voice recognition — reconnaissance vocale.