Originally posted by: Viper GTS
It's actually not that unfeasible.
Look at how long the recordings were, with that kind of sample you could tabulate each identical keystroke & compare it to distributions of characters in language.
My dad used to impress us when were kids by having us make up a letter substitution code & write him a letter. He would then break it.
A 10 minute recording would give a MUCH larger sample size, & with the aid of a PC the potential is even greater.
Viper GTS
Yeah but who takes 10 minutes to type a password? The ability for this program to correctly identify a password is slim, especially if that password isn't an actual word, and uses numbers. The program would need a primer, it would need to be knowledgable about the person typing, and that is very unlikely if the microphone is "planted".
Also figuring that not everyone types the same, and the sound made by a key would probably vary depending on with what finger it was hit and what direction it was hit from. I'm wondering if they used a group of people to make the samples and not just one person. I'm assuming the program also monitors the time in milliseconds between keystrokes to pick up patterns that way, but again many people type differently. I'm wondering how accurate the program would be if the person typing was using "Hunt-and-Peck". Also, if the microphone is indeed planted, it would've been much easier for the planter to install an on-wire keylogger, which is MUCH MUCH more accurate.
Facebook
Twitter