Most modern cameras have built-in microphones that are sufficient to capture high-quality audio for analytics purposes.
Video analytics has grabbed the attention of physical security customers in recent years. The introduction of AI systems that can recognize what it sees has helped improve not just security but also business operations. According to Allied Market Research, the global AI video surveillance market size was valued at US $14.83 billion in 2020 and is projected to reach $55.22 billion by 2030, at a CAGR of 14.9 percent during the period.
A related area that doesn’t get as much attention as it deserves is audio analytics. Just as video content can provide valuable insights based on spatial and physical attributes, audio content can provide clues to threats and issues. For instance, detecting sounds like screams can help security officers know if there is a problem.
Most modern cameras have built-in microphones that are sufficient to capture high-quality audio for analytics purposes. Many customers don’t make use of these microphones because of regulatory concerns in certain areas. But the advantage of audio analytics is that the system doesn’t need to record sounds continuously.
How it works
Sound detection software utilizes an intricate blend of parameters, ranging from decibel levels to the energy in various frequencies over time, to identify sounds. These applications are designed to overlook routine ambient noise, such as traffic, conversations, and music, and instead trigger an alert upon recognizing a precise acoustic signature. This reduces the likelihood of false positive or false negative outcomes, even in demanding settings such as train stations and prison cells.
In a
blog post on the website of Axis Communications, Derek van der Vorst, Founder of Sound Intelligence, explains that the program only operates in buffer mode, recording a few seconds before and after the detection to allow security to verify the sound and preserve it for forensic evidence.
Audio analytics becomes even more helpful when embedded on edge, in network cameras, speakers, access control systems, etc., offering several advantages. Edge processing eliminates the need for a central server, lowers bandwidth consumption, and enables better real-time security.
Why your customer may need audio analytics
Audio analytic software can detect several kinds of sounds, providing customers with a range of functions that can assist in better decision-making. Because of machine learning, the potential of the system to acquire knowledge of sounds is massive – its limitation is only the processing power.
There are several situations in which audio analytics could come in handy. Here are a few:
Gunshot detection
Gunshot detection has received much interest recently after an unprecedented increase in gun violence in the US. According to Gun Violence Archive, more than 600 mass shootings have occurred in the last three years. That’s almost two a day on average.
Gunshot detection solutions can identify sounds from a number of firearms, including handguns, shotguns, rifles, and automatic weapons. Once detected, the system alerts security personnel, who can make use of video surveillance cameras for instant verification.
This is useful in a number of verticals but perhaps most notably in public places like malls, train stations, and airports where the identity of every person cannot be verified. Unfortunately, school shootings have also become a major issue in countries like the US, even when entry to campuses is restricted.
Detect aggressive behavior
Many instances of physical assault or altercation begin with an aggressive verbal exchange. Detecting this can help security officers prevent a situation from escalating. The analytics software looks for sound patterns associated with duress, anger, or fear and instantly alerts the relevant personnel.
This feature can come in handy in public and private areas where people gather. An assault can happen anywhere at any time. In fact, this is potentially one of the most useful features in many settings to protect people and assets. Imagine someone screams in a public place - our first reaction is to check if everything is ok. Making surveillance devices intelligent enough to have the same response is the way forward.
Glass break noise detection
The sound of glass breaking can indicate several things, but from a security perspective, it is often someone breaking in through a window that we are worried about. Many buildings consider motion sensors at windows to detect any untoward activity. Using audio analytics is more cost-efficient.
Any vertical where buildings must have perimeter protection can make use of this feature. A bank, for instance, needs to make sure that no one breaks in through a window.
Detect car alarms
As the name says, the car alarm detection feature detects the sound of car alarms and alerts security. This can be of great value in parking lots, garages, and any place where the number of cars is too many for people to monitor manually.
Other useful features
The features mentioned above are just a few among many. Some companies are even working on offering detection of a baby crying, dog barking, etc. Going forward, we may see more customization in this area, depending on the verticals of the application.
For instance, the sounds of an oil refinery would be different from that of a retail store. Audio analytics can improve situational awareness because security operators can now receive alerts even when the visuals may not indicate a problem.