MAPS
##################################################################################
__ __ _____ _____
| \/ | /\ | __ \ / ____|
| \ / | / \ | |__) | (___
| |\/| | / /\ \ | ___/ \___ \
| | | |/ ____ \| | ____) |
|_| |_/_/ \_\_| |_____/
For
Mouse Audio Process Streaming
Scenario interaction mouse/audio
Hugo Baranger aka HV$VH & Alexandre Sune aka Mr_H4l3x
##################################################################################
#1 The general idea:
On a computer, sound is generated in association with the mouse 'click' event.
Click is the trace of the human interaction with the machine.
If we can record in a database all the clicks done on a specific computer we can produce
an history of the human interactions.
This database of interactions is used to generate an audio composition.
#2 How to do that:
A/ Capture each "click" events, and record it in a database.
The record would be formated like this:
ID | DAY | MONTH | YEAR | Hour | Minute | second | positionX | position Y
To capture each "click" we can use Xbindkey, XDOTOOL and a bash script associated.
The datas are recorded in a SQL database with the same script.
B/ Generate audio.
The database is dumped in a file,
and each line of the file is interpreted by a PureData Patch to produce audio synthesis.
At the end of the file the database is dumped and the content of the file is replaced
by a new and fresh one.
This will run forever.
C/ Stream the audio composition in realtime.
The computer will run has a webserver and use icecast2 to perform Live Streaming.
The Puredata Patch will use the element "MP3cast~" from the Library "Unauthorized"
to connect audio of the Computer to the Listening Port of Icecast2.
#3 the project is divided in 3 Blocs.
Bloc #1 capture mouse event and record it
Bloc #2 interprete recorded event to produce audio content
Bloc #3 Stream it Live Baby!
-----------------
| BLOC #1 |
-----------------
At the very beginning i was thinking about « XDOTOOL » to manage this part, with a little bash script and « voilà! »… But, after many try I came to the conclusion that this method slow down all the computer’s running process, because the script have to look at « click » events and record these in the same time, hum, not so easy to do… but keyloggers exist so there will be a way to do it, maybe a sneeky one.
The Python way:
Let me introduce you « pynput » a library that allows you to control and monitor input devices.
Here is the script used:
#import the library for listening inputs and date/time
from pynput.mouse import Listener
import datetime
#function used when a « click » is detected
#1 check the date/time && get the mouse position (x y)
#make a string with theses informations
#record it in a text file
def is_clicked(x, y, button, pressed):
if pressed:
d=datetime.datetime.now()
dd=d.strftime("%d %m %Y %H %M %S")
finalString=format(dd)+" "+format(x)+" "+format(y)+"\n"
theFile=open("/home/pauline/theMouseRecord.txt","a", encoding="utf-8")
theFile.write(finalString)
theFile.close()
#a simple listener to trigger the clicked mouse
with Listener(on_click=is_clicked) as listener:
listener.join()
Now we just have to launch the script at startup and it will be good :) (it's done from PureData with the "shell" object, more info bellow)
Ok, stage completed ✅
-----------------
| BLOC #2 |
-----------------
To make some noise, we need to use a versatile programmation language: PudeData.
I will let Hugo explain more about the sound research and work, and for now i just want to explain the very basis of the patch. We use PureData vanilla and some useful library:
"ggee" to get the "shell" object (ability to send bash commands directly from PureData)
"Unauthorized" to get the "MP3cast~" object (ability to stream audio via icecast2 from PureData)
"Zexy" to have access to a nice collections oaf objects (very useful for Hugo)
We need to read a text in PureData line by line and extract informations (date/time/x/y), so I used to launch a simple bash script from PureData using "shell" object.
The bash script:
#!/bin/bash
input="path/to/the/file.txt"
while IFS= read -r line
do
echo "$line"
sleep 1s
done < "$input"
Once launched the script will read and output the txt line by line and at the end the "shell" object will notice us with a "bang" like this we can re-launch the bash script when the PureData process is done.
Ok, PureData can read a txt file ✅
PureData can play sounds now ... (insert Hugo part here)
Dear PureData, You can read, you can sing, can you stream?
For audio the streaming process we need the "MP3cast~" object correctly configured (portnumber/password) on localhost. With Icecast2 running...
We need another bloc.
-----------------
| BLOC #3 |
-----------------
Icecast2 is a very nice solution to stream audio on the web, we use it with Hugo for more than a year on another project (a generative web radio "ASH.b-22" based on number radios). Icecast is very stable and reliable, but it's a pain in the ass to configure SSL on it. Because Icecast need a specific file made with "fullchain.pem" and "privkey.pem"
You can do it like this:
sudo cat /etc/letsencrypt/live/YOUR-DOMAIN-NAME/fullchain.pem /etc/letsencrypt/live/YOUR-DOMAIN-NAME/privkey.pem > /etc/icecast2/bundle.pem
Once it's done, you need to add a specific port for SSL in the Icecast.xml:
<listen-socket>
<port>8080</port>
</listen-socket>
<listen-socket>
<port>8443</port>
<ssl>1</ssl>
</listen-socket>
And add the path of the SSL bundle:
<ssl-certificate>/etc/icecast2/bundle.pem</ssl-certificate>
We keep the no-SSL one still open, because PureData's "MP3cast~" object can't stream on ssl (the stream from PureData has a localhost target so it's not so problematic 😬 ) maybe it will be useful to make this modification on "MP3cast~" with the sources... ;)
Once the Icecast2 server is running, we need to add an audio element linked to the stream in the no-VNC webpage of SUDA.
And "voilà!" 😅
H4l3x