Decoding DMR radio using RTLSDR, GQRX and Pulseaudio

I recently watched Melissa Elliot’s talk from DEFCON 21 titled Noise Floor Exploring Unintentional Radio Emissions 2013 which inspired me to buy a Realtek SDR (software defined radio) dongle from eBay.

I am running Ubuntu (13.04 to be specific).

I am assuming that you already have your SDR dongle working and gqrx installed on your system.

Sources#

I have compiled these instructions from a couple of sources:

• GQRX SDR Google Group thread: Using gqrx with Virtual pulseaudio source?
• Kersle.net blog: Redirect Audio Out to Mic In (Linux)

DSD#

DSD is the digital speech decoder that you need to pass the audio output from your software radio (gqrx in our case) through in order to get useful audio output on the other side.

The DSD Homepage links to the GitHub repos for dsd and mbelib, as well as some Windows downloads.

1. Install cmake, pavucontrol, and the sndfile library (if not installed):

sudo apt-get install cmake libsndfile1-dev pavucontrol

2. Install mbelib from https://github.com/szechyjs/mbelib:

cd ~/code
git clone https://github.com/szechyjs/mbelib.git
cd mbelib
mkdir build
cd build
cmake ..
make -j4
sudo make install

3. Install dsd from https://github.com/szechyjs/dsd.git:

cd ~/code
git clone https://github.com/szechyjs/dsd.git
cd dsd
mkdir build
cd build
cmake ..
make -j4
sudo make install

Pipe sound from GQRX into DSD using Pulseaudio volume control#

I have compiled these directions from this entry on kirsle.net.

dsd does not have Pulseaudio support, so we will create a Pulseaudio sink that emulates OSS, which dsd does support using padsp.

Using pavucontrol, we will redirect the sound output from gqrx to the recording input that dsd is listening on.

I have been looking in the 440MHz - 460MHz range in my city and actually came across some analog voice, but we are more interested in the digital speech traffic. I believe this is the frequency range of the Motorola digital radios.

1. Open pavucontrol.

2. Start padsp and dsd using the /dev/dsp OSS emulated Pulseaudio sink:

padsp -- dsd -i /dev/dsp -o /dev/dsp -fa -ma

Your console output will probably look something like this:

% padsp -- dsd -i /dev/dsp -o /dev/dsp
Digital Speech Decoder 1.7.0-dev (build:v1.6.0-26-ge40e89a)
mbelib version 1.2.5
Audio In/Out Device: /dev/dsp

3. Start the DSP playback in gqrx by clicking the power button in the toolbar or selecting Start DSP in the File menu.

4. At this point, you should see an entry for GNU Radio: SDR and OSS Emulation[dsd]: Audio Stream on the Playback tab in the Pulseaudio Volume Control window and an entry for OSS Emulation[dsd]: Audio Stream in the Recording tab. We will be manipulating the sources to the right of the audio sink titles.

5. Tune the software radio receiver frequency to a station that looks like it might be a digital station. I am using a Normal width filter and the Narrow FM mode in the Receiver Options tab of gqrx. The waveform for my target station looks like this:

6. In the Input Devices tab, select Monitors from the Show: dropdown. This will display the automatically created Pulseaudio sinks which monitor the sound output of each of your sound devices. Make sure that the monitor sink for your main sound output is unmuted.

7. Switch to the Recording tab and select the Monitor of [main sound output sink] from the source to the left of the OSS Emulation[dsd]: Audio Stream from title. This pipes the audio output from your main sound output sink into the OSS emulated DSP sink created by padsp that dsd is listening on.

8. As soon as you tune the software receiver to a compatible station, dsd should lock on to the digital sound stream and start decoding. You will probably see output like this streaming by in the terminal where dsd is running:

Sync:  +DMR       mod: GFSK inlvl:  2%  slot0  [slot1]  TLC
Sync:  +DMR       mod: GFSK inlvl:  2% [slot0]  slot1   CSBK
Sync:  +DMR       mod: GFSK inlvl:  2%  slot0  [slot1]  VOICE Header

9. Have fun!

Additional Tricks#

My computer has Pulseaudio outputs for the main speaker plug, as well as the HDMI output. In order to clear my speaker outputs of the raw, un-decoded radio sound, I am piping the audio output of gqrx to the HDMI sink instead of the main speaker output sink. My dsd OSS emulation sink is then set to listen on the HDMI monitor sink.