Voice 'Fingerprints' Change Crime-Solving
The FBI is trying to develop a system that could make your voice as unique and recognizable as your fingerprint. Although not yet at its peak potential, the technology currently helps investigators with tasks such as verifying Osama bin Laden videos and locating gunshots.
Running an audio clip of someone's voice through the system, called FASR, prompts bright squiggly lines to rise and fall. Each voice is quite distinct. While the pattern isn't as definitive as DNA, the FBI says, FASR gets pretty close.
When a new Osama bin Laden audio or video tape pops up on the Internet, the forensic analysts at the FBI Audio Lab in Quantico, Va., process the voice through this system. The results allow them report whether the tape is authentic.
It doesn't matter what language is being spoken, Steven Lanser who heads up the FBI audio team says. Regardless of whether one is speaking Arabic, Urdu or English, a voice follows a particular pattern.
"We use an automated system and that system looks at a variety of different voice features that it extracts automatically," Lanser says. "Then it gives us a confidence score of the speaker versus an unknown speaker."
In other words, does this Osama bin Laden sound like that Osama bin Laden?
Lanser says the FASR system science is not perfect. By itself, a match on FASR isn't enough to convict anyone in a court of law. Added to other pieces of evidence, however, it could provide intelligence to either include or eliminate a suspect.
The new world of audio forensics helps with more traditional crime-fighting as well. Gunshot-detection systems are the next frontier in audio forensics, some law enforcement officials say. Cities across the country are starting to put noise sensors on rooftops. The system analyzes the time it takes for the sound of gunfire to travel to various sensors and creates a picture of where the gun went off using GPS, providing key information to police within seconds.
Minneapolis Police Chief Tim Dolan has used the technology, called ShotSpotter, in two high-crime areas for several years. In a recent case, Dolan says, a 16-year-old girl was caught in gang-related crossfire with police. The suspects claimed the officers fired first.
"We were able to tell from the ShotSpotter exactly how many shots were fired and were able to break down who shot when," he says. "It made a big difference and was able to take away an alibi from the people involved in that shooting.
The technology provides a street address for where shots were fired. And if there is more than one shot, it can determine which direction the person firing the shots was going.
While Dolan says the system has been helpful in Minnesota, the technology isn't perfect. Ice popping or the crack of roofing guns could set off the sensors. Consequently, in the beginning there were many false alarms. A lot of those problems went away when Minneapolis started coordinating its gun-detection systems with cameras, Dolan says.
"We're getting fewer false positives and it is getting better," he says. "What's out here today is much better than what was out there three years ago."
Other local law enforcement agencies appear to agree: More than 20 American cities have installed new gunshot-detection systems.
Copyright 2023 NPR. To see more, visit https://www.npr.org.