The Sound of Science

Eastern professor’s sound-based gunshot detection system could potentially save lives

News reports of school shootings motivated EMU faculty member Tareq Khan to invent a small, affordable, and effective gunshot detection system.

The Sound of Science

Eastern professor’s sound-based gunshot detection system could potentially save lives

Fire-related fatalities in schools within the United States are quite rare. On average, fire departments respond to about 3,200 school fires annually, with just one death, according to the National Fire Protection Association.

Compare those numbers to firearm-related deaths in schools.

Since 2018, there have been 226 school shootings in the U.S., resulting in 142 deaths—an average of 20 people killed per incident—according to statistics compiled by Education Week. Data from the gun violence prevention advocacy group Sandy Hook Promise shows that more than 390,000 students have experienced gun violence at school since the tragic mass shooting at Columbine High School in 1999.

One can trace the low number of fire-related fatalities in schools to the prevalence of smoke detectors and fire-suppression systems. Most of these installations came in the wake of tragic incidents such as a school fire in Chicago that killed 92 children and three teachers in 1958.

Imagine if every classroom had a device that could immediately detect the sound of gunfire and report critical information to local law enforcement. Consider the number of lives that could be saved by something as affordable and easy to install as a smoke detector.

This isn’t just a dream. The device exists and is currently in the research and development stage at Eastern Michigan University.

Smart sensor

The Deep Learning-Based Gunshot Detection IoT System is the brainchild of Tareq Khan, EMU associate professor of electrical and computer engineering and computer and electrical engineering technology.

“I have not personally been exposed to any type of school shooting, but seeing those incidents on the news really bothers me,” says Khan, who began developing the system in 2023. Earlier this year, he published a paper detailing his research in a special issue of the journal Advances in IoT and Machine Learning for Smart Homes. “This is an area where my academic expertise and the needs of society overlap.”

Khan’s gunshot detection device is about the size of a typical power brick used to charge a laptop. The device can be plugged into a standard electrical outlet or hardwired into an existing electrical system.

The device uses Artificial Intelligence to analyze sound waves and identify gun shots in real time.

“We recorded gunshots using two types of blank guns: a pistol and a fully automatic rifle,” Khan says. “Then we created a deep learning model that classifies sounds the device detects as either ‘gunshot’ or ‘non-gunshot.’ Our initial tests showed the device can detect gunfire within a 40-foot range. It can also distinguish between actual gunfire and the sound of gunfire in a movie or TV soundtrack, as well as other loud sounds like thunder or a balloon pop.”

The audio-based device can detect gunshots within at least 40 feet.

“During an active shooter situation, the window for providing life-saving care is small. First responders will tell you that every second matters. This automated reporting system will save time and save lives.”

— Tareq Khan

Khan’s system includes a data notification component that wirelessly transmits GPS location information to emergency responders via the Internet.

“Within a second of gunfire detection, the system sends an alert to emergency responders with exact map coordinates,” Khan says. “During an active shooter situation, the window for providing life-saving care is small. First responders will tell you that every second matters. This automated reporting system will save time and save lives.”

Khan’s device is intended for indoor use. Besides schools, other applications could include shopping malls, banks, places of worship, movie theatres—even private homes.

SpotShotter—a variation of a sound-based gunfire detection system—is in place in cities such as Chicago, Seattle, and Cleveland. However, it’s designed only for outdoor use, and it has been criticized for being costly, inaccurate, and ineffective. Chicago recently chose not to renew its SpotShotter contract.

In 2023, Eastern installed an AI-based gun detection video analytics system called ZeroEyes, which is integrated into existing exterior digital security cameras on campus. Khan’s audio-based device could potentially complement that system.

“There is no way to hide sounds,” Khan says. “When a gun is fired, sound waves spread in all directions.”

Mohammed Hagras (right), a junior electrical and computer engineering major from Cairo, Egypt, is Khan’s project assistant.

Targeted testing

To test the gunshot detection system and solicit feedback from law enforcement, Khan enlisted the help of the City of Dearborn Police Department and its Police Chief Issa Shahin (BS97, MS16).

At the department’s gun range last March, Khan and Mohammed Hagras—a junior electrical and computer engineering major at Eastern from Cairo, Egypt—logged data from the device during several rounds of live gunfire.

“Officers used several types of rifles, shotguns, and handguns,” Khan says. “The device detected 100 percent of the shots. It was a very exciting moment for us. Real guns are much louder than blank guns, which means the device’s range of detection for actual gunfire could be much higher than 40 feet.”

Chief Shahin was equally impressed.

“I think Professor Khan’s system holds great promise for any community, school, workplace or public space looking to increase its safety measures,” Shahin says. “When officers respond to a shooting, they must quickly identify the shooter’s exact location, which can be difficult. Anything that can help officers do that in real time is helpful. Time equates to lives saved. God forbid, if a shooting does happen, you want to have the right training and equipment in place. This device has the potential to be part of that equation.”

The device has undergone numerous lab and field tests in advance of possible production.

Eastern has filed for a provisional U.S. patent on the gunshot detection system. Khan says five public schools within the Ann Arbor, Ypsilanti, and Dearborn districts have expressed interest in the system, and he hopes to open a startup company soon to begin production.

“We all want kids and teachers to be safe while in school,” Khan says. “The school principals I’ve talked with say they often feel they work in a state of fear. We need to provide a solution for this security problem in our public places. The gunshot detection system could be a piece that helps solve that puzzle.”

By Jeff Samoray

Photos by Charlotte Smith