This is a collection of command-line and GUI tools for capturing and analyzing audio data.
The most interesting tool is called keytap - it can guess pressed keyboard keys only by analyzing the audio captured from the computer's microphone.
Check this blog post for more details:
Keytap: description and some random thoughts
Video: short demo of Keytap in action
The keytap2 tool is another interesting tool for recovering text from audio. It does not require training data - instead it uses statistical information about the frequencies of the letters and n-grams in the English language. A more detailed description of the tool is available here: Keytap2 discussion
Video: short demo of Keytap2 in action
CTF: can you guess the text being typed?
"This works incredibly well.
I hope you realize what you've created (and made available to every person in the world)." -- ffpip
"I just tried it and it works incredibly well. It kind of makes me want to stop using a mechanical keyboard." -- Karawebnetwork
"This attack and Van Eck phreaking are why Edward Snowden, while typing passwords and other sensitive information, would pull a blanket over himself and his laptop." -- aarchi
"This is what mechanical keyboard users deserve" -- super guy
"fuck.." -- Lluis Franco
Dependencies:
-
SDL2 - used to capture audio and to open GUI windows libsdl
[Ubuntu] $ sudo apt install libsdl2-dev [Mac OS with brew] $ brew install sdl2 [MSYS2] $ pacman -S git cmake make mingw-w64-x86_64-dlfcn mingw-w64-x86_64-gcc mingw-w64-x86_64-SDL2
-
FFTW3 (optional) - some of the helper tools perform Fourier transformations fftw
Linux, FreeBSD, Mac OS, Windows (MSYS2 + MinGW)
git clone https://github.com/CheKecha19/kbd-audio
cd kbd-audio
git submodule update --init
mkdir build && cd build
cmake ..
make
Short summary of the available tools. If the status of the tool is not stable, expect problems and non-optimal results.
Name | Type | Status |
---|---|---|
record | text | stable |
record-full | text | stable |
play | text | stable |
play-full | text | stable |
view-gui | gui | stable |
view-full-gui | gui | stable |
keytap | text | stable |
keytap-gui | gui | stable |
keytap2 | text | development |
keytap2-gui | gui | stable |
- | extra | - |
guess_qp | text | experiment |
guess_qp2 | text | experiment |
key_detector | text | experiment |
scale | text | experiment |
subreak | text | experiment |
key_average_gui | gui | experiment |
-
record-full
Record audio to a raw binary file on disk
./record-full output.kbd [-cN]
-
play-full
Playback a recording captured via the record-full tool
./play-full input.kbd [-pN]
-
record
Record audio only while typing. Useful for collecting training data for keytap
./record output.kbd [-cN] [-CN]
-
play
Playback a recording created via the record tool
./play input.kbd [-pN]
-
keytap
Detect pressed keys via microphone audio capture in real-time. Uses training data captured via the record tool.
./keytap input0.kbd [input1.kbd] [input2.kbd] ... [-cN] [-CN] [-pF] [-tF]
-
keytap-gui
Detect pressed keys via microphone audio capture in real-time. Uses training data captured via the record tool. GUI version.
./keytap-gui input0.kbd [input1.kbd] [input2.kbd] ... [-cN] [-CN]
**Live demo (WebAssembly threads required) **
-
keytap2-gui record.kbd n-gram-dir [-pN] [-cN] [-CN]
Detect pressed keys via microphone audio capture. Uses statistical information (n-gram frequencies) about the language. No training data is required. The 'record.kbd' input file has to be generated via the record-full tool and contains the audio data that will be analyzed. The 'n-gram-dir' folder file has to contain n-gram probability files for the corresponding language.
./keytap2-gui record.kbd ../data
-
view-full-gui
Visualize waveforms recorded with the record-full tool. Can also playback the audio data.
./view-full-gui input.kbd [-pN]
-
view-gui
Visualize training data recorded with the record tool. Can also playback the audio data.
./view-gui input.kbd [-pN]
Any feedback about the performance of the tools is highly appreciated. Please drop a comment here.