September 5, 2014 -
Last month a team of biometric researchers at the West Virginia University traveled to Twinsburg, Ohio’s Twins Days Festival, the world’s largest annual gathering of twins and multiples.
At the event, the WVU researchers introduced audio and video data collection to their process.
“Automated voice recognition is emerging as a biometric method,” said Jeremy Dawson, research assistant professor in the Lane Department of Computer Science and Electrical Engineering. “There has been little-to-no research performed on how similar twins’ voices are”.
In order to ensure the collection of quality data, researchers set up portable sound isolation rooms on site that protected their research data from almost all background noise.
“These are much more efficient collection wise. All you could hear was the participant,” said Casey Norville, a senior biometric systems and computer engineering major who manned the station. “This was a really cool hands-on, real world application of biometric science.”
The event provides a unique opportunity for biometric researchers to collect a large amount of data from a rare sample of the population.
Student researchers, who typically collect fingerprints, facial photos, and audio and video recordings indoors at the WVU lab, brought these lab setups outdoors.
For 10 hours a day, these students must make quick adjustments to their setup to account for the shifting conditions.
“Twins Days allows us to build large datasets that contain things such as facial images that would be considered challenging in typical biometric applications and scenarios,” said Dawson.
Students say that the most challenging component of Twins Days was expediting the collection process and maintaining quality data.
Read more about the project on the WVU website.