This guide is heavily based on derek's guide but i used liquidsoap instead of darkice and vnc for headless linux server.

Lets say you have some users talking in a mumble server and want to brodcast that conversation but dont want people to join and interfare but want people able to listen to. So, here's how to do that.

Requirements:

• Mumble Client
• PulseAudio
• Savonet Liquidsoap
• Icecast

Installation

Here i'm gonna do it in ubuntu.

1

apt-get install mumble pulseaudio liquidsoap liquidsoap-plugin-lame

If you are doing this in a headless server you'll also need vncserver. Realvnc is my favorite but tightvnc will also do.

1

apt-get tightvncserver

Now install icecast

1

apt-get install icecast2

Configuration

Icecast

if you choose not to configure icecast2 during installation here's how to do that manually.

Open /etc/icecast2/icecast.xml

1

nano /etc/icecast2/icecast.xml

There is many configuration with comments, change them if you want, but at least we need to change the default passwords. change <source-password>, <relay-password> and <admin-password>

Now enable icecast2, open /etc/default/icecast2

1

nano /etc/default/icecast2

1

ENABLE=true

Now start icecast2

1

/etc/init.d/icecast2 start

Point your browser to SERVER_IP:PORT to see icecast.

Vnc

Skip this if you already have installed and access to Desktop Environment.

Now to run mumble, we need a DE. you may install lxde (as a lightweight DE)

1

apt-get install lxde

Start a vnc session which will also start the X server.

1

vncserver

Now go to your own computer's desktop and using any VNC client, connect to you server. Credentials will be like host: serverip:1 and your vnc password.

Pulseaudio

Assuming you were using ssh so far, now switch to DE, and start lxterminal to further complete tasks ahead.

Now we need to create a virtual device in pulseaudio from where we'll be capturing the mumble session and to where mumble should send its output. To do that, run,

1

pactl load-module module-null-sink sink_name=stream

Mumble

Start mumble from menu or running mumble in terminal. Most important thing to set is

• Settings->Audio Output System = PulseAudio Device = Null Output

You can leave other configs as default.

Now connect to desired mumble server/channel.

Liquidsoap

Now the final task, link mumble's output to icecast. It is a good idea to check out quick start guide which'll give you a clear idea about Liquidsoap.

We'll create a liq script.

1

nano sk.liq

Now the main tricky part, you need to change the content of the script.

1
2

#!/usr/bin/liquidsoap
output.icecast(%mp3,     host = "example_host", port = 8000,      password = "source_password", mount = "gittu",     input.pulseaudio(device = "stream.monitor"))

You need to change example_host to your icecast server's ip or address. source_password to the <source-password> set in icecast.

Now everything is set, fire things up

1
2

chmod +x sk.liq
./sk.liq

Now go to icecast and listen your mumble conversation

Toubleshooting

1. If icecast fails to start, check that the port (default 8000) is not being used by another process. It took away 30 minutes of my life, nginx was blocking 8000 port.
2. Try running liquidsoap in verbose mode liquidsoap -v sk.liq if you are having trouble with it.

Thats all Let me know your thoughts in comment section.