Stop Moaning or Get Filtered, Researchers Use Sound to Identify Porn

Stop Moaning or Get Filtered, Researchers Use Sound to Identify Porn

comments
0
By Richard Johnson on May 25, 2011

In another yet another coup for technologists, researchers out of the Korea Advanced Institute of Science and Technology in Daejeon, South Korea have developed software that can identify porn just from the moaning and grunting. Software to visually identify porn has existed for years, but the researchers decided to take advantage of another sense and are now able to successfully identify porn 93% of the time.

Using an ingenious method, researchers sampled sound clips of just half a second long and created spectrograms for each clip. They then compared those clips against test clips achieving incredible results. Like any first try, this particular porn stopper method isn’t perfect. There were a number of false positives which stemmed from rigorous crowds, loud noises, cheers and cries, consequently marking them porn even though the video clips were safe for work. It’s understandable that loud noises, cheering, and even crying could confuse the system into thinking it had found some naughty bits, all of those noises are routine with myself and bedfellows.

While the method is ground breaking, itʼs a bit less efficient than visual filters which need just a single frame of a personʼs fleshy parts to decide if they are doing the NSFW dance. As it turns out, computers aren’t nearly as adept at identifying the difference between a steamy scene and a foggy morning when compared to a dude with sex on the brain. But by merging visual clues like perky breasts with auditory clues like moaning, scientists hope that their system will get it wrong a whole lot less.

Itʼs hard to say if scientists are pursuing ground breaking research or just trying to find a way to watch porn while at work, but hereʼs hoping that they ultimately fail miserably so we can still have a chance of catching a flash of an ass and part of a moan while weʼre stuck behind those pesky at work firewalls.

Please login to submit comments