Low memory requirement (the demo instance runs on a Raspberry Pi3B+, using less than 10% of total memory even when transcoding)
User management
Recommendation engine
Audio transcode for maximum interoperability and low bandwith requirements
Persistent play queue across sessions
Subsonic API
Album artist
Multi-value tags: artists, genres, ...
Custom tags (ex: mood, genre, albummood, albumgrouping, ...)
MusicBrainzID support to handle duplicated artist and release names
Playlists, (only using Subsonic API for now)
Starred Album/Artist/Tracks (only using Subsonic API for now)
Systemd integration

Recommendation engine

LMS provides several ways to help you find the music you like:

Tag-based filters (ex: Rock, Metal and Aggressive, Electronic and Relaxed, ...)
Recommendations for similar artists and albums
Radio mode
Searches in album, artist and track names
Most played/Recently added music

The recommendation engine makes use of Self-Organizing Maps.
Please note this engine:

may require some significant computation time on large collections (ex: half a hour for 40k tracks)
makes use of computed data available on AcousticBrainz. Therefore your music files must contain the MusicBrainz Identifier for the recommendation engine to work properly (otherwise, only tag-based recommendations are provided)

Subsonic API

The API version implemented is 1.12.0 and has been tested on Android using the official application, Ultrasonic and DSub.

Since LMS uses metadata tags to organize music, a compatibility mode is used to navigate through the collection using the directory browsing commands.

The Subsonic API is enabled by default.

Note: since LMS stores hashed and salted passwords, it cannot handle the token authentication method defined from 1.13.0.

Installation

Note: this installation process and the default values of the configuration file and the systemd service file have been written for Debian Stretch. Therefore, you may have to adapt commands and/or paths in order to fit to your distribution.

Build dependencies

apt-get install g++ autoconf automake libboost-filesystem-dev libboost-system-dev libavcodec-dev libavutil-dev libavformat-dev libav-tools libmagick++-dev libpstreams-dev libconfig++-dev libpstreams-dev ffmpeg libtag1-dev

You also need W4, that is not packaged yet on Debian. See installation instructions.

Build

Get the latest stable release and build it:

git clone https://github.com/epoupon/lms.git lms
cd lms
autoreconf -vfi
mkdir build
cd build
../configure --prefix=/usr

configure will report any missing library.

Note: in order to customize the installation directories, you can use the following options of the configure script:

--prefix (defaults to /usr/local).
--bindir (defaults to $PREFIX/bin).

make

Note: you can use make -jN to speed up compilation time (N is the number of compilation workers to spawn).

Raspberry 3 build notes

Due to memory limitations, you may need to build Wt4 in Release mode if you want to compile it natively on a Raspberry Pi3B+.

Note: the build process is really long, roughly 1 hour to build Wt4 + LMS.

Deployment

Note: the commands of this section require root privileges.

make install

Create a dedicated system user:

useradd --system lms

Copy the configuration files:

cp /usr/share/lms/default.conf /etc/lms.conf
cp /usr/share/lms/default.service /lib/systemd/system/lms.service

Create working directories and give access to the lms user:

mkdir /var/lms
touch /var/log/lms.log
touch /var/log/lms.access.log
chown lms:lms /var/lms /var/log/lms.log /var/log/lms.access.log

Note: don't forget to give the lms user read access to the music directory you want to scan.

Configuration

LMS uses a configuration file, installed by default in /etc/lms.conf. It is recommended to edit this file and change relevant settings (listen address, listen port, working directory, Subsonic API activation, ...).

All other settings are set using the web interface (user management, scan settings, transcode settings, ...).

Reverse proxy settings

LMS is shipped with an embedded web server, but it is recommended to deploy behind a reverse proxy. You have to set the behind-reverse-proxy option to true in the lms.conf configuration file.

Here is an example to make LMS properly work on myserver.org using nginx.

server {
    listen 80;

    server_name myserver.org;

    access_log            /var/log/nginx/myserver.access.log;

    proxy_request_buffering off;
    proxy_buffering off;
    proxy_buffer_size 4k;

    location / {

      proxy_set_header        Host $host;
      proxy_set_header        X-Real-IP $remote_addr;
      proxy_set_header        X-Forwarded-For $proxy_add_x_forwarded_for;
      proxy_set_header        X-Forwarded-Proto $scheme;

      proxy_pass          http://localhost:5082;
      proxy_read_timeout  120;
    }
}

Run

systemctl start lms

Two log files are used:

/var/log/lms.log: logs of the lms daemon
/var/log/lms.access.log: access logs of the embedded web server

To connect to LMS, just open your favorite browser and go to http://localhost:5082

To make LMS run automatically during startup:

systemctl enable lms

Upgrade

To upgrade LMS, you need to update the master branch and rebuild/install it:

cd build
git pull
make

Then using root privileges:

make install
systemctl lms restart

Read the release notes to check if any relevant setting has been added into the configuration file.

Any change in the database schema is made automatically by the application during startup.

Credits

Wt (http://www.webtoolkit.eu/)
bootstrap3 (http://getbootstrap.com/)
ffmpeg project (https://ffmpeg.org/)
Magick++ (http://www.imagemagick.org/Magick++/)
MetaBrainz (https://metabrainz.org/)
Bootstrap Notify: https://github.com/mouse0270/bootstrap-notify