There’s a man who stands in a room at ideaSpace shouting. It’s part of the job. Occasionally he’ll get someone to smash windows, fire guns, anything that sounds like trouble, he’s interested.
Dr Chris Mitchell is CEO of Audio Analytic, the company he founded in 2008 to add a second sense to video-based security systems, hearing.
Audio Analytic produces software that automatically recognises sounds through computer analysis, alerting security teams to otherwise mute pictures that don’t necessary relay the entire picture. Here’s their startup story.
There’s no escaping video security monitoring, most commonly in the form of closed circuit television (CCTV); it’s everywhere, at work, on the street, even in the home. One estimate claims 1.85 million CCTV cameras in the UK alone, but according to Dr Mitchell most of these systems have fundamental limitations.
They need a direct line of view for detection, are often affected by lighting levels and are not capable of detecting certain situations such as a person being verbally aggressive.
Audio, an untapped domain says Dr Mitchell and one he believes the current range of professional security products simply do not use to detect events that could not be detected by video feeds alone.
The technology is used to recognise sounds indicative of distress, adversity or crime such as screaming, vocal aggression, gunshots or break-ins. By recognising these sounds and bringing them to the attention of security operators, suitable assistance can be dispatched to deal with the event that is unfolding.
At its heart is ‘Core Logger’, a product with a patent pending that is added to security products such as CCTV cameras or video recorders to recognise specific types of sound. When combined with ‘Sound Packs’, it provides manufacturers with the ability to detect sounds of interest in a diverse range of noisy environments.
The idea grew out of Dr Mitchell’s PhD research into automatically classifying different genres of music such as classical, jazz and pop. After finishing his PhD he embarked on a Kauffman Fellowship in the United States and started searching for applications for the technique as well as further developing the technology.
During the fellowship he worked with Cisco systems and quickly realised that the same technology could be applied to the security industry where he found that sound was simply not being used.
“The security market, in common with many other markets, is currently moving from traditional analogue solutions to digital networks allowing for more data than simply a video feed to be utilised,” says Dr Mitchell. “This move towards IP networks is driving high growth in the security market.”
In relation to video detection alone, Audio Analytic’s products do not depend on those problems already identified, lighting levels, a direct line of sight, verbal aggression.
Other technologies working with audio also fall short says Dr Mitchell. “At the moment some security equipment monitors the volume of sound, and alerts security personnel whenever the volume is elevated - this suffers from many false alerts (e.g., post coming through the letter box) and in practice is unusable for most problems.”
A number of other companies have designed systems to recognise very specific sounds - especially gunshots - but cannot manage the range of sounds that Audio Analytic’s technology is capable of in a single system says Dr Mitchell.
THE BUSINESS MODEL
Audio Analytic licenses its technology to security equipment manufacturers who integrate it into their products, along with a simple interface that allows non-expert users to quickly and effectively start using the company's products. This strategy is designed to help Audio Analytic become widely available to end users.
"Revenue is typically generated on a per-unit-sold basis, and profits by generating more revenue that we expend," Dr Mitchell said.
Following a couple of seed stage investments and research grants, Audio Analytic completed a Series A funding round with Cambridge Angels in October 2010.
blog comments powered by Disqus