A new acoustic algorithm creating "smart ears" in the wall can isolate conversations between two people in a crowded hall.
By JUDY SIEGEL-ITZKOVICH
Criminals often meet in crowded rooms to prevent eavesdropping by law enforcement officials; even if a microphone picks up their voices, it is difficult to separate them from other conversations taking place at the same time.
But now a new acoustic algorithm creating "smart ears" in the wall - developed by engineer Dr. Sharon Gannot at Bar-Ilan University in Ramat Gan - can isolate conversations between two people in a crowded hall. There are other important uses for this, such as in hearing aids.
"The new algorithm uses a number of microphones dispersed in the given space," explains Gannot. "First we locate the persons whose conversation interests us by scanning the space. The information received by the microphone setup is then screened using a special device we have developed which succeeds in blocking out the unwanted conversations. By doing so, we have, for all practical purposes, managed to overcome 'the cocktail-party problem' of isolating one specific conversation from among many taking place simultaneously," he says.
"Now, when the police want to eavesdrop on suspects such as drug dealers, who frequently meet in crowded locations, it will be possible to listen in if several microphones are planted in the room," explains Gannot. He adds that BIU researchers and developers are currently working to improve the algorithm. In future, they hope to be able to track interlocutors who are moving about in a room. An acoustic laboratory is being established in the School of Engineering - the only one of its kind in Israel - which will allow experiments under real-life conditions.
"In the next stage, it will be possible to assimilate the algorithm into hearing aids," adds Gannot. "For example, it will be possible to enable people to listen to a conversation taking place in the direction in which they are looking."
GENES AT THE BALLOT BOX
The decision to vote - but not whom you vote for - is partly genetic, according to a new study published in the American Political Science Review. The University of California research, by Drs. James Fowler, Christopher Dawes and Laura Baker, is the first to show that genes influence a wide range of political activities. In the July issue of the Journal of Politics, the team showed that individuals with a variant of the MAOA gene are significantly more likely to have voted in the 2000 presidential election. Their research also demonstrates a connection between a variant of the 5HTT gene and voter turnout, which is moderated by religious attendance.
Their initial research is based on voter turnout records in Los Angeles matched to a registry of identical and non-identical twins. These comparisons show that identical twins are significantly more similar in their voting behavior than fraternal twins. The results indicate that 53% of the variation in voter turnout is due to differences in genes. The results also suggest that, contrary to decades of conventional wisdom, family upbringing may have little effect on participatory behavior.
To replicate these findings, the researchers went beyond the California voter data to examine patterns nationwide using the National Longitudinal Study of Adolescent Health conducted from 1994 to 2002. Such data have been used in a wide variety of genetic studies, but this is the first time the data have been used to show that participatory political behavior is heritable.
"These findings are extremely important for how we think about political behavior," said Fowler. For example, it is widely known that parents and children exhibit similar voting behavior, but this has always been interpreted as learned behavior. It is also well-known that these particular genes influence social behavior, but it has not been widely appreciated that social behavior plays an important role in political behavior. We are not robots - the genes just seem to make it more likely that some of us will get involved in politics."