-----------------------------------------------------------------------------------
Directory Structure
-----------------------------------------------------------------------------------
key_board_sidechannel_10307933_10307936
├── audio_to_file.py ├── createwmat.py
├── dict
├── dispsim.py
├── final_ver5.py
├── Readme
├── recordwords.sh
├── simlist.py
└── sorcmd.sh
-----------------------------------------------------------------------------------
Dependencies
-----------------------------------------------------------------------------------
Numpy and Scipy --- For matrix and array operations
Matplotlib --- For plotting
Gnuradio --- For basic audio processing and recording (any other program can be used to directly record from mic )
Python 2.x --- Aliased as python2 everywhere in the program
-----------------------------------------------------------------------------------
Repository
-----------------------------------------------------------------------------------
Entire code along with the audio recordings can be found in the github repository. Audio recordings are not included in this submission archive and should be downloaded from the repository as the file sizes are large.
HTTP link: https://github.com/zeroXzero/acoustic_crypt or clone the entire project by git clone https://github.com/zeroXzero/acoustic_crypt.git
git clone git://github.com/zeroXzero/acoustic_crypt.git
Audio recordings can be found in anechoic_chamber_recordings folder
in the repository.
-----------------------------------------------------------------------------------
Program files
-----------------------------------------------------------------------------------
Readme --- This readme file
dict --- Sample dictionary containing 27K words
simlist.py --- ADJ, DIST and NEAR tables for QWERTY keyboard
createwmat.py --- Program that creates constraint matrices (weq,wadj etc)
audio_to_file.py --- Used for recording keystrokes from standard audio input recordwords.sh --- Used for continuous recording of various words
sorcmd.sh --- Sorts the dictionary according to word length and adds header which makes further accessing easier. Makes use of createwmat.py to create constraint matrices.
dispsim.py --- Displays original constraints for any given word
final_ver5.py --- Original program that implements the attack
-----------------------------------------------------------------------------------
Usage and brief explanation
-----------------------------------------------------------------------------------
1) Audio recording -- optional python2 audio_to_file.py -I hw:0,0 -N 48000 sampledump
check -h option for more details
2) Sorting dictionary and creating constraint matrices for dictionary words ./sorcmd.sh dict
This created a 'dict_sorted' file and 'wmat' directory that contains the constraint matrices for various wordlengths
3) Actual constraints
python2 dispsim.py <word>
e.g python2 dispsim.py dear
This program can be used for generating the actual constraints for a word
4) Final attack
python2 final_ver5.py <audio_recording>
e.g. python2 final_ver5.py anechoic_chamber_recordings/laptop_keybrd/<recording>
-----------------------------------------------------------------------------------