AURA Engine
Aura Engine is a play-out engine as part of Automated Radio (Aura) system, specifically build for the requirements of community radios.
Features
Play tracks from multiple sources
It's possible to air playlists with music or recordings stored on the filessystem, via external streams or live from a studio.
Blank Detenction / Silence Detecter
The engine offers a simple way to detect scenarios where no music is on air. It possible to configure the sensitivity of the Silence Detector and automatically transition play-out to a Fallback Playlist (see Auto Pilot).
Auto Pilot
In case there is no schedule delivered by Steering, engine provides multiple fallback handling scenarios. The available fallbacks are evaluated in following order:
- Timeslot Fallback: //TODO explain
- Show Fallback: //TODO explain
- Station Fallback: //TODO explain
Dynamically switching of source channels
TODO extend: * switch the soundserver at the correct time to a given source for a specific show
Multiple Recorders
TODO extend: * record what is broadcasted
Stream to an Icecast Server
TODO extend: * stream to an icecast server
Multichannel Line-out
TODO extend: * play to line-out
Architecture
AURA Engine as part of the AURA Radio Suite uses an modulear architecture based on a REST API. All external information is retrieved using JSON data-structures.
Required Data Sources
The AURA Project "Dashboard" provides the GUI to organize shows, schedules/timelsots and organize uploads in form of playlists. Those playlists can be organized in timeslots using a fancy calendar interface.
These data-sources need to be configurated in the "engine.ini" configuration file:
# The URL to get the Calendar via PV/Steering
calendarurl="http://localhost:8000/api/v1/playout"
# The URL to get show details via PV/Steering
api_show_url="http://localhost:8000/api/v1/shows/${ID}/"
The AURA Project "Tank" on the other hand delivers information on the tracks, playlists to be played and its meta-data:
# The URL to get playlist details via Tank
importerurl="http://localhost:8040/api/v1/shows/${SLUG}/playlists"
More information you can find here: https://gitlab.servus.at/autoradio/meta/blob/master/api-definition.md
Provided API Endpoints
Soundserverstate: Returns true and false values of the internal In- and Outputs
/api/v1/soundserver_state
Trackservice:
/api/v1/trackservice/<selected_date>
/api/v1/trackservice/
Installation
Hardware Requirements
This depends on how many audio sources and targets you are going to use, but for the most common scenarios any current hardware should be sufficient. For the audio devices it is required to use an interface which has supported ALSA drivers.
Aura Engine is tested with following audio devices
- ASUS Xonar DGX,
- Roland Duo-Capture Ex
- Onboard Soundcard (HDA Intel ALC262)
- Native Instruments Komplete Audio 6
Software Requirements
Operating System: Any linux system with ALSA, PulseAudio or Jack2 support should work. It is tested and coded on Debian Stretch and Ubuntu 18.0 with Python 3.6+.
Getting Started
git clone https://gitlab.servus.at/autoradio/engine
Install System Packages
On a Debian / Ubuntu machine:
sudo apt install \
git \
python3 python3-pip \
redis-server \
liquidsoap liquidsoap-plugin-icecast \
mariadb-server libmariadbclient-dev \
quelcom \
liquidsoap-plugin-alsa liquidsoap-plugin-pulseaudio
File Formats: Depending on what stream you are going to send, and what recordings you are going to use:
sudo apt install \
liquidsoap-plugin-aac # for aac support
liquidsoap-plugin-flac # for flac support
liquidsoap-plugin-lame liquidsoap-plugin-mad # for mp3 support
liquidsoap-plugin-opus # for opus support
liquidsoap-plugin-vorbis # for ogg support
To simply install support for all available file formats do:
sudo apt install \
liquidsoap-plugin-all
Install Python Packages
sudo pip3 install -r requirements.txt
Setup Database
mysql -u root -p
CREATE DATABASE aura_engine CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
CREATE USER 'aura'@'localhost' IDENTIFIED BY 'secure-password';
GRANT ALL PRIVILEGES ON aura_engine.* TO 'aura'@'localhost';
Alternative Sound Servers
Beside ALSA the sound servers Jack Audio and Pulse Audio are supported.
Using JACK:
Install the JACK daemon and GUI:
sudo apt-get install jackd qjackctl
Please ensure to enable "realtime process priority" when installing JACK to keep latency low. Now, you are able to configure your hardware settings using following command:
qjackctl
Next you need to install the JACK plugin for Liquidsoap:
sudo apt install \
liquidsoap-plugin-jack
Configuration
Run
sh init.sh
This creates the folder /var/audio and copies some default configuration to /etc/aura/engine.ini
After that, you have to edit the settings in /etc/aura/engine.ini. Ensure to take your time to carefully review those settings!
Hardware
Soundcard
Hard/Soft
When you use ALSA, you will have to play around with ALSA settings. In the folder ./modules/liquidsoap is a scipt called alsa_settings_tester.liq. You can start it with 'liquidsoap -v --debug alsa_settings_tester.liq'. Changing and playing with settings may help you to find correct ALSA settings.
Line In
You can configure up to five line ins. Your hardware should support that. When you use JACK, you will see the additional elements popping up when viewing your connections (with e.g. Patchage).
Recordings
You can configure up to five recorders. You find the settings in the main config file engine.ini. You can choose between different output formats.
Streams
You can configure up to five streams. You find the settings in the engine.ini. You can choose between different streaming formats.
Running the Engine
To start the AuRa Engine execute:
systemctl start aura-lqs
systemctl start aura-engine
and on system boot run following:
systemctl enable aura-lqs
systemctl enable aura-engine
The first service starts the LiquidSoap Engine, while the latter boots the actual AuRa Engine.
Logging
You can access the service logs using:
journalctl -u aura-lqs
and
journalctl -u aura-engine
respectively.
Development
To run the LiquidSoap code during develpment execute: ./run.sh
Components
aura.py: It is the server which is connected to the external programme source (e.g. aura steering and tank), to liquidsoap and is listening for redis pubsub messages. This precious little server is telling liquidsoap what to play and when.
Guru: The commandline tool for interacting with the server. Also provides the communication from Liquidsoap to the Python (Command-)Server.
Liquidsoap: The heart of AuRa Engine. It uses the built in mixer, to switch between different sources. It records everything and streams everything depending on your settings in aura.ini.
Frequently Asked Questions
ALSA Settings
In the Liquidsoap Logs I get 'Error when starting output output_lineout_0: Failure("Error while setting open_pcm: Device or resource busy")!'. What does it mean?
You probably have set a wrong or occupied device ID.
How can I find the audio device IDs, required for settings in engine.ini?
-
ALSA: You can get the device numbers or IDs by executing:
cat /proc/asound/cards
-
Pulse Audio: You might not need this for Pulse Audio, but still, to see all available devices use:
pactl list
If you cannot find correct ALSA settings
Well, this is - at least for me - a hard one. I could not manage to find correct ALSA settings for the above mentioned soundcards. The best experience i had with the ASUS Xonar DGX, but still very problematic (especially the first couple of minutes after starting liquidsoap). Since i enabled JACK support i only use that. It is also a bit of trial and error, but works pretty much out of the box.
If you experience 'hangs' or other artefacts on the output signal
-
reduce the quality (especially, when hangs are on the stream) or
-
install the realtime kernel with
apt install linux-image-rt-amd64 reboot
or
-
invest in better hardware
Resources
- Python: https://docs.python.org/
- Redis: https://redis.io/
- Liquidsoap: https://www.liquidsoap.info/doc-1.4.0/
- Jack Audio: https://jackaudio.org/