Audio Analytic uses advanced psychoacoustic modelling to produce software which automatically recognises sounds – such as smoke alarms, breaking glass or baby cries – that is used by leading original equipment manufacturers (OEMs) around the world.
The company’s origins are in founder Chris Mitchell’s self-funded PhD in sound information systems and signal processing at Anglia Ruskin University, in Cambridge, in which he used computers for music genre recognition through tone, pitch and timbre. Chris admits that he loves “boxes that make noises” and analysing those noises appealed as a PhD subject. And, conveniently, his research lent itself to potential commercial applications.
Chris says: “As I was getting towards the end of the PhD, I started to see how it could be applied and saw there wasn’t anybody applying this type of technology in sound recognition.”
Chris was awarded a fellowship to investigate the technology developed as part of his PhD. He received training at Harvard Business School and worked with Cisco Systems in San Jose, California. It was in the United States that Chris had his Eureka moment – the idea of using sounds from CCTV systems to detect incidents and crimes that could not be identified from the video feed alone.
The funding, the products and licensing the technology
Chris founded Audio Analytic in 2008 and received a £15k proof of concept grant from the East of England Development Authority (EEDA) in August 2008. This was followed by a £20k grant from NESTA/NCGE in January 2009. Commercial trials of its first product for the market started that April.
Around this time Chris was at a meeting in Cambridge that led to an introduction to Stew McTavish, director of the then recently-launched ideaSpace, who explained what it was and his vision for it. It was enough to get Chris to join as one of the earliest members and he says he benefitted from the community and connections from which ideaSpace members benefit.
Chris says: “Coming from outside of Cambridge University, it gave us access to the University. We had assistance from the Institute for Manufacturing (IfM) and other University departments.
“There is a lot of knowledge available at ideaSpace and you draw from it – there were a lot of useful meetings as well as random chats with members that turned into useful information. There is value in chatting with people who are doing similar things to you.”
Audio Analytic took a major step forward in October 2010 when it closed a funding round led by Cambridge Angels, allowing it to invest in product development, delivery and marketing.
At the heart of the company’s business is the ability to detect the acoustic signature of a sound. Audio Analytic has developed a system that reduces thousands of comparable sounds into their core constituent parts and summarises them into what the company calls a ‘software sensor’. These can represent the summary of thousands of hours of audio recordings of a particular sound like baby cries, breaking glass, smoke alarms and car alarms. The sensors are lightweight and fit into the most compact specifications of existing hardware designs. The technology was developed from years of cutting-edge research.
Norway-based Zenitel AS, a global supplier of security communication systems, announced a licensing agreement in November 2011 to integrate Audio Analytic’s sound recognition technology with its communication systems. Further licensing agreements followed in September 2012 with Next Level Security Systems and AxxonSoft.
Expanding the applications for new markets
Audio Analytic’s early focus was on professional security applications for its software sensors to monitor schools and colleges, hospitals and civic buildings, shops and banks, transport and urban centres. The technology helps to alert CCTV operators to car thefts or break-ins (car alarm detection), building break-ins (glass break detection), verbal and physical violence in buildings or transport systems (aggression detection) and firearms crimes (gunshot detection).
In April 2014, the firm raised a further £900,000 to help fund its expansion into the very competitive and fast-changing home security/automation and consumer electronics market place.
Property protection is enhanced by Audio Analytic’s software sensors which detect breaking glass, smoke alarms and carbon monoxide alarms. The sensors can be used to trigger text messages, emails or video recording and transmission when incidents occur. The Keywords software sensor detects the specific sound associated with a predefined command or emergency word (e.g. Help, Help, Help!).
The Baby Cry software sensor has been designed to detect a range of cries from babies to toddlers aged up to 24 months. The system can detect up to 10 meters from a sound source using a typical MEMS microphone. The technology can be used to trigger responses such as audio monitoring, video recording and transmission, and increased playback volume to make sure an audio event is not missed. All of this helps provide greater peace of mind for parents.
Leading the way in sound recognition
From its inception, the company’s core values were founded in being the emerging leader in sound recognition – and those values continue to guide its development. It is constantly seeking new applications that will offer greater peace of mind to anyone who chooses its technology.
Audio Analytic was based at ideaSpace West and later at ideaSpace City for about a year before moving to its own offices in St Andrew’s Street, Cambridge, in May 2014.
Chris says: “We moved from West to City to have more space – then we outgrew that and moved here.”
Audio Analytic increased in size to 10 staff in December 2014 and it has high-profile customers worldwide in the home automation and security sectors, including Cisco, Swann and Zenitel.
The company has won industry recognition and attracted media attention. The firm was named as one of the 16 most exciting start-ups in Cambridge by The Observer, a leading UK Sunday newspaper, in December 2013. It was shortlisted for the Three New Things competition, organised by Virgin Media, from almost 100 entrants, and it was a semi-finalist for Richard Branson’s Pitch to Rich competition in 2014 to find the UK’s best start up.
The company won the Consumer Electronics Technology section of the Institution of Engineering and Technology (IET) Innovation Awards in November 2014 for its sound recognition technology for OEMs (smoke alarm, glass break, baby cry, aggression detection).
Recalling the role ideaSpace has played in the company’s development, Chris acknowledges the help of Stew and the management team, adding: “We had access to the ideaSpace community and all that it entails.”
And his advice for current members on how to make the most of what ideaSpace offers?
“Make sure you are always talking to people, even if you have no reason to! Go out of your way to talk to them.”
For more on Audio Analytic see www.audioanalytic.com