Researchers from the University of Portsmouth have designed artificial intelligence software to allow CCTV cameras fitted with microphones to recognise sounds such as crowd noise and windows smashing.
The cameras would 'hear' noises associated with violence, and automatically swivel to record the event.
A three-year surveillance study funded by the Engineering and Physical Sciences Research Council hopes to speed up the capture of criminal acts on camera and improve overall response times.
The system could even be used to recognise specific words, according to Dr David Brown, director of the university's Institute of Industrial Research.
"The longer artificial intelligence is in the software the more it learns," he said.
Dr Brown also addressed concerns of a 'Big Brother' society and the overuse of CCTV, claiming that the research project was not aimed at snooping on private conversations.
"We are only listening for specific words associated with violence, not full conversations," he said.
The announcement comes amid growing concern about the use of CCTV by councils to tackle 'trivial' offences such as littering and dog fouling.
Sir Simon Milton, head of the Local Government Association, warned on Monday that local councils run the risk of alienating the public and could lose the right to use CCTV-based surveillance methods altogether.
Back in January Westminster Council ended its trial of attaching microphones to CCTV cameras and confirmed that it would not roll out the technology across the region after outcries from privacy campaigners.
And, yep, it'll run Android rather than RiscOS
US engineering giant's cost-cutting outsourcing plan is on the rocks, according to insiders
HP Envy X2 laptop only affordable if you've got loadsamoney
Counterfeit code-signing certificates enabling hackers to hide malware being sold by cyber criminals
Certificates can be used as part of layered obfuscation to evade detection by anti-virus software