Difference between revisions of "Music Player Daemon/Tips and tricks"

From ArchWiki
Jump to navigation Jump to search
m (No more rc.conf, spaces, no more repo references)
m (mpdas)
Line 21: Line 21:
{{bc|# nano /etc/mpdasrc}}
{{bc|# nano /etc/mpdasrc}}
* Edit the file {{ic|~/.mpdscribble/mpdscribble.conf}}:
* Edit the file {{ic|/etc/mpdasrc}}:
username = LASTFM USER
username = LASTFM USER

Revision as of 09:07, 5 October 2013

Go back to Music Player Daemon.

Tips and tricks


MPD does not provide a tool to preserve the file structure of your audio files. Check out beets.

Last.fm scrobbling

To scrobble your songs to Last.fm when using MPD, there are several possibilities.


mpdas is an AudioScrobbler client for MPD written in C++. It uses curl and libmpd. Actually mpdas is the best way to send song updates to Lastfm, because it supports the latest AudioScrobbler protocol (2.0). Also, in case of downtime or connection problems mpdas will cache the played songs to ~/.mpdascache.

mpdasAUR is available in the AUR. Visit the official website for more information.

After you have instaled mpdas, do the following (as root):

# nano /etc/mpdasrc
  • Edit the file /etc/mpdasrc:
username = LASTFM USER
password = PASSWORD IN MD5
debug = 1
#runas is an optional feature. It change the user mpdas runs as
runas = USERNAME
#host is an optional feature to set the current mpd host
host = HOST
#mpdpassword is an optional feature to set the current mpd password
mpdpassword = MPDPASS
#port is an optional feature to set the current mpd port
port = MPDPORT

To get your password in md5 hash, just run:

echo -n 'PASSWORD' | md5sum | cut -f 1 -d " "
  • Add mpdas to the file in which you start mpd as well (e.g. ~/.xinitrc):
pidof mpdas >& /dev/null
if [ $? -ne 0 ]; then
 mpdas &


mpdscribble is a daemon available in the official repositories (if you prefer, mpdscribble-gitAUR is available in the AUR). This is arguably the best alternative, because it is the semi-official MPD scrobbler and uses the new "idle" feature in MPD for more accurate scrobbling. Also, you do not need root access to configure it, because it does not need any changes to /etc at all. Visit the official website for more information.

After you have installed mpdscribble, do the following (not as root):

$ mkdir ~/.mpdscribble
  • Create the file ~/.mpdscribble/mpdscribble.conf and add the following:
 host = your mpd host # optional, defaults to $MPD_HOST or localhost
 port = your mpd port # optional, defaults to $MPD_PORT or 6600
 log = /home/YOUR_USERNAME/.mpdscribble/mpdscribble.log
 verbose = 2
 sleep = 1
 musicdir = your music directory
 proxy = your proxy # optional, e. g. http://your.proxy:8080, defaults to none

 # last.fm section, comment if you do not use last.fm
 url = http://post.audioscrobbler.com/
 username = your last.fm username
 password = your last.fm password
 journal = /home/YOUR_USERNAME/.mpdscribble/lastfm.journal

 # libre.fm section, comment if you do not use libre.fm
 url = http://turtle.libre.fm/
 username = your libre.fm username
 password = your libre.fm password
 journal = /home/YOUR_USERNAME/.mpdscribble/librefm.journal

Please note that passwords can also be written down as MD5:

echo -n 'PASSWORD' | md5sum | cut -f 1 -d " "
  • Add mpdscribble to the file in which you start mpd as well (e.g. ~/.xinitrc):
pidof mpdscribble >& /dev/null
if [ $? -ne 0 ]; then
 mpdscribble &

mpdscribble per user with systemd

Since systemd is default on new installations, starting mpdscribble with .xinitrc is outdated. mpdscribble should be started per user with systemd. See Mpd#Start_MPD_per_user_with_systemd and Systemd/User.

Install mpdscribble.

Create a directory for mpdscribble and copy the example .conf

$ mkdir ~/.mpdscribble
$ cp /usr/share/mpdscribble/mpdscribble.conf.example ~/.mpdscribble.conf

With the current version mpdscribble 0.22-5, adding the sections mentiond above to ~/.mpdscribble.conf is not necessary. Just add your username and password.

Then, edit a new file ~/.config/systemd/user/mpdscribble.service

Description = MPD Scribbler

ExecStart = /usr/bin/mpdscribble -D
Restart = always

WantedBy = default.target

Now, enable and start mpdscribble.service as per user

$ systemctl --user enable mpdscribble
$ systemctl --user start mpdscribble

Sonata and Ario

Sonata has built-in support for scrobbling, although that requires the program to run the whole time. Additionally, Sonata does not cache the songs if they cannot be forwarded to Last.fm at the time of playing, meaning they will not be added to the statistics.


The daemon lastfmsubmitd is a daemon which may be installed from the official repositories as well. To install it, first edit /etc/lastfmsubmitd.conf to reflect your requirements and activate both lastfmsubmitd and lastmp systemd services.

Last.fm playback

Native Last.fm playback

In January 2013 Last.fm changed their protocols and the maintainer of mpd decided not to fix it

Never play on start

This feature is present in mpd after version 0.16.2. When this feature is enabled, the mpd process will always start in the "paused" state, even if a song was playing when mpd was stopped. Add the line below to your mpd.conf file to enable this feature.

restore_paused "yes"

Example configuration: Output with 44.1 KHz at e. g. 16 bit depth, multiple programs at once

Why these formats? Because they are standard CDA, because ALSA on its own allows more than one program "to sound" only with dmix — whose resampling algorithm is inferior — and because dmix by default resamples anything lower to 48 KHz (or whatever higher format is playing at the time). Also, some get clicking sounds if at least mpd.conf is not changed this way.

What is the downside? These settings cause everything (if necessary) to be resampled to this format, such as material from DVD or TV which usually is at 48 KHz. But there is no known way to have ALSA dynamically change the format, and particularly if you listen to far more CDs than anything else the occasional 48 → 44.1 is not too great a loss.

The following assumes that there are not already other settings which conflict resp. overwrite it. This applies especially to the current user's potential ~/.asoundrc — which MPD as its own user ignores, therefore the following should go to /etc/asound.conf:

defaults.pcm.dmix.rate 44100 # Force 44.1 KHz
defaults.pcm.dmix.format S16_LE # Force 16 bits
audio_output {
        type                    "alsa" # Use the ALSA output plugin.
	name			"HDA Intel" # Can be called anything or nothing tmk, but must be present.
        options                 "dev=dmixer"
        device                  "plug:dmix" # Both lines cause MPD to output to dmix.
	format	        	"44100:16:2" # the actual format
	auto_resample		"no" # This bypasses ALSA's own algorithms, which generally are inferior. See below how to choose a different one.
	use_mmap		"yes" # Minor speed improvement, should work with all modern cards.

samplerate_converter		"0" # MPD's best, most CPU intensive algorithm. See 'man mpd.conf' for others — for anything other than the poorest "internal", libsamplerate must be installed.
Note: MPD gives the mp3 format a special treatment at decoding: it is always put out as 24 bit. (The conversion as forced by the format line only comes after that.)

If one wants to leave the bit depth decision to ALSA resp. MPD, comment out resp. omit the dmix.format line and change the one for mpd with format to "44100:*:2".

Note: Crossfading between files decoded at two different bit depths (say, one mp3 and one 16 bit flac) does not work unless conversion is active.

Control MPD with lirc

There are already some clients designed for communications between lircd and MPD, however, as far as the practical use, they are not very useful since their functions are limited.

It is recommended to use mpc with irexec. mpc is a command line player which only sends the command to MPD and exits immediately, which is perfect for irexec, the command runner included in lirc. What irexec does is that it runs a specified command once received a remote control button.

First of all, please setup your remotes as referred to the Lirc article.

Edit your favored lirc startup configuration file, default location is ~/.lircrc.

Fill the file with the following pattern:

     prog = irexec
     button = <button_name>
     config = <command_to_run>
     repeat = <0 or 1>

An useful example:

## irexec
     prog = irexec
     button = play_pause
     config = mpc toggle
     repeat = 0

     prog = irexec
     button = stop
     config = mpc stop
     repeat = 0
     prog = irexec
     button = previous
     config = mpc prev
     repeat = 0
     prog = irexec
     button = next
     config = mpc next
     repeat = 0
     prog = irexec
     button = volup
     config = mpc volume +2
     repeat = 1
     prog = irexec
     button = voldown
     config = mpc volume -2
     repeat = 1
     prog = irexec
     button = pbc
     config = mpc random
     repeat = 0
     prog = irexec
     button = pdvd
     config = mpc update
     repeat = 0
     prog = irexec
     button = right
     config = mpc seek +00:00:05
     repeat = 0
     prog = irexec
     button = left
     config = mpc seek -00:00:05
     repeat = 0
     prog = irexec
     button = up
     config = mpc seek +1%
     repeat = 0
     prog = irexec
     button = down
     config = mpc seek -1%
     repeat = 0

There are more functions for mpc, run man mpc for more info.

Control MPD with phone

remuco via Bluetooth

  • Install remucoAUR from the AUR - a wireless remote control for several Linux media players (http://remuco.sourceforge.net/index.php/Remuco).
  • Transfer remuco client -- jar/jad files from /usr/share/remuco/client/ to your phone and install it.
  • Run remuco-mpd (as current user).
  • Run remuco on your phone, define a new bluetooth remuco connection (pair first if you have not done this previously) and explore its capabilities.


There are several MPD-specific clients available in the app-store/market of Symbian, Android and iOS.

MPD and PulseAudio

mpd supports both local and remote use of PulseAudio.

Local (as your own user)

No special options are required; just add a pulse output as described in the comments of mpd's config file.

If you want to start mpd with systemd, put the following file into /etc/systemd/system/mpd-myuser.service:

.include /usr/lib/systemd/system/mpd.service

Description=Music Player Daemon (running as MYUSER)


(replacing MYUSER for your username). Then do sudo systemctl enable mpd-myuser && sudo systemctl start mpd-myuser.

Local (with separate mpd user)

When run as its own user as per the wiki instructions, mpd will be unable to send sound to another user's pulseaudio server. Rather than setting up pulseaudio as a system-wide daemon, a practice strongly discouraged by upstream, you can instead configure mpd to use pulseaudio's tcp module to send sound to localhost:

First, uncomment the tcp module in /etc/pulse/default.pa or $XDG_CONFIG_HOME/default.pa (typically ~/.config/pulse/default.pa) and set as an allowed IP address; the home directory takes precedence:

### Network access (may be configured with paprefs, so leave this commented
### here if you plan to use paprefs)
#load-module module-esound-protocol-tcp
load-module module-native-protocol-tcp auth-ip-acl=
#load-module module-zeroconf-publish

Additional IP ranges in cidr notation may be added using ; as the separator. Once this is complete, restart pulseaudio:

$ pulseaudio --kill
$ pulseaudio --start -or- start-pulseaudio-x11/kde

Next, edit /etc/mpd.conf and add a new pulse output pointing to as a "remote" server:

audio_output {
       type		"pulse"
       name		"Local Music Player Daemon"
       server		""

Once this is added, restart mpd.

Enable the output in mpd; you should now have a working local mpd, usable by all users whose pulseaudio servers allow sound from


As with any PulseAudio-enabled program, mpd can send sound over the network. The complete PulseAudio system is not required on the server running mpd; libpulse is the only requirement to act as a source and is already a dependency of mpd.

In order to send audio from mpd to another computer follow the directions above, editing /etc/mpd.conf on the server running mpd using the IP address of the target computer and /etc/pulse/default.pa or $XDG_CONFIG_HOME/default.pa (typically ~/.config/pulse/default.pa) on the target computer using the IP address of the server.

Once this is done, the server's mpd source should show up on the target computer while playing or paused as a normal source able to be rerouted and controlled as usual; there will be no visible source on the target while mpd is stopped.

Cue Files

No additional steps are needed for cue support in mpd since 0.17. MPD has its own integrated parser which works with both external and embedded cuesheets. For example, the command mpc load albumx/x.cue loads the file music_directory/albumx/x.cue as playlist; or in the case of an CUESHEET tag, mpc load albumx/x.flac.

HTTP Streaming

Since version 0.15 there is a built-in HTTP streaming daemon/server that comes with MPD. To activate this server simply set it as output device in mpd.conf:

audio_output {    
	type		"httpd"    
	name		"My HTTP Stream"
	encoder		"vorbis"		# optional, vorbis or lame
	port		"8000"
#	quality		"5.0"			# do not define if bitrate is defined
	bitrate		"128"			# do not define if quality is defined
	format		"44100:16:1"

Then to listen to this stream simply open the URL of your mpd server (along with the specified port) in your favorite music player. Note: You may have to specify the file format of the stream using an appropriate file extension in the URL. For example, using Winamp 5.5, You would use rather than

To use mpd to connect to the stream from another computer.

mpc add