From ArchWiki
Revision as of 18:00, 9 July 2009 by Svoufff (talk | contribs) (Building a performance template)
Jump to: navigation, search


Linuxsampler is a software audio sampler.It can be loaded with libraries in the gigasampler format (.gig). A benefit of LS compared to other samplers available for linux, is its streaming from hard disk capabilities, allowing the usage of huge libraries (several GBs..). By now, "Linuxsampler" refers to a group of software interacting together :

  • the sampler engine itself : linuxsampler
  • a gig file editor : gigedit
  • several graphical frontends : qsampler (QT based), jsampler-classic and jsampler-fantasia (both java based)


IMPORTANT NOTE 1 : Follow the installation order of theses components to get a working set of apps.

IMPORTANT NOTE 2 : All packages are available from AUR.The whole set can be grabbed as standard packages or cvs versions. Standard packages in AUR are up-to-date but the sources tarballs released by the LS team can be old. CVS versions are bleeding edge.They appear to work well in i686 and x86_64 architectures but it's never sure that all packages will build against each other at any time.

You should choose between a full set of standard packages OR cvs packages but do not mix them. Whatever you choose, but particularly with cvs packages, you should keep in a safe place a set of binary packages that work together.

The backend : Linuxsampler

It is the "engine" of the sampler, which performs all the heavy and time critical computational tasks of handling MIDI events, calculating the audio data and sending the final audio data to your sound card(s).LinuxSampler itself usually runs as own process in the background of the computer and usually does not show up anything on the screen, or at most it can be launched to show status informations and debug messages in a console window.Usually you will also need a frontend to control it.

Build and install :


and check that everything is ok :

$ linuxsampler

then CTRL-c to quit.

The editor : Gigedit

Gigedit allows you to edit and create instruments for the Gigasampler format, which can be used with LinuxSampler as well as with Tascam's Gigastudio.You can use gigedit as stand-alone application without Linuxsampler running or in live-mode, started from any frontend ("edit" button).In this case, all your modifications are audible in realtime.

Build and install (optional, you can skip this step if you don't need the editor):


and check that everything is ok :

$ gigedit

The editor should pop-up in "stand-alone" mode as indicated in the down-right corner. Close the window.

The frontend : Qsampler

This frontend provides the user a set of menus, buttons, sliders, dials, etc. to allow the user to control the sampler in a convenient way.It does not perform any signal processing tasks, so you can see it as a "face" of the sampler.From Qsampler you can load/unload .gig files into Linuxsampler and set their midi channel number, volume, jack output etc...and finally store all this setup in a file for a speedy re-opening of your "sampling orchestra".

Build and install :


Building a default template in qsampler

Here, I will use jack-keyboard as a midi controller, but everything below applies as well with a master keyboard or midi interface providing midi output.

  • Start jack then jack-keyboard then qsampler
  • Let's define some audio outputs in qsampler :
Go to menu -> View -> Devices (or F11 shortcut or the green pci card icon in the toolbar).
On the left panel, click on "Audio Devices".Now we can see the audio drivers available in the right panel.
Choose "Jack" in the Driver menu.The default config should give you a pair of audio outputs at a sample rate of 44100.
Click on the Create button down the window.

Note : In some systems, qsampler crashes if this window is closed or even backgrounded while the Bottom-Right panel is still in view.Easy workaround : click on "Audio devices" or "Midi Devices" on the left panel before leaving or unfocusing the window.

  • If you want LS to auto-connect to jack system outputs when you launch your template :
On the left panel, click on "Audio Jack Device 0" , then on right panel select "Audio Jack 0" from the "Channel" drop-down list.
Choose an output for the JACK_BINDINGS parameter, like "system:playback 1".
Select "Audio Jack 1" from the "Channel" drop-down list and assign JACK_BINDINGS parameter value, like system:playback 2"
  • Now let's create a midi input also :
On the left panel, click on "Midi Devices".
On the right panel, choose "jack" (jack-midi) in the Driver menu.
Hit the create button.
  • If you want LS to auto-connect to jack-keyboard midi output when you launch your template :
Select the "Midi jack device" port 0" on the left panel.
On the right panel, for the "JACK_BINDINGS" parameter value, choose "jack-keyboard:midi out".

Note that if jack-midi is not enabled in Qjackctl, you won't find the midi client in the drop-down list.You can check in Qjackctl setup : "Midi driver" should be set to "seq".A restart of jack and qsampler is needed.

Also, if jack_keyboard or whatever midi controller you defined in here is not started before qsampler starts, you will likely receive an error from qsampler not finding the midi client.In that case, you can still start the midi device and connect it through qjackctl.

  • Finally close the Devices Config window and save your template.

TIP : You can define as much midi ports as midi controllers you use.The same applies to audio outputs, e.g for sending separated instrument outputs for recording in Ardour.

Building a performance template

  • Open qsampler and load your configuration template.
  • Press the "Add channel" button in the toolbar (Ctrl-A).A "channel" in the LS terminology is an object (visually a strip) wich stores the settings for a given instrument : Name and location of the gig file, inputs & outputs used, midi channel, volume, solo/mute status, midi cc applied etc..much like a track in a sequencer.
  • The i/o settings ot the channel pops up.
    • Choose a gig file
    • Choose an instrument.A gig file can contain different "instruments".Basically a gig file is composed of a pool of sounds and some configuration possibilities of theses samples named "instruments".For example one insrument can load only the forte samples and apply some adsr on it, while another one can load every samples in the pool, and organize them as layers that will be triggered by the velocity applied

on the controller.This way, very different types of of sound can be obtained from the same pool of samples, wich is more efficient in terms of sound-design and disk space...

    • Set the midi and audio i/o for the channel from the different drop-down lists.All the options you will find here were created earlier in the device configuration setup stored in your config template.If you would like to change some of it, you can also do it from here with the "midi input setup" and "audio output setup" buttons.
    • Press ok.
  • The channel is added to the main window and the gig file is loaded into ram.Well, only the start of the sounds used in the instrument are cached in memory, the main task of LS will be to stream the biggest part of them directly from disk.That's how it's possible to load 30 GBs of samples with only 1 GB of ram.
  • You should now be able to play your instrument from jack-keyboard or whatever controller is tied to LS.

If you get no sound, check the volume sliders both on the channel strip and on the main toolbar.

  • Repeat these steps as many times as you wish to complete your "sampling orchestra".
  • Save your template under a new name so you keep your default template clean if you want to start another orchestra from ground.


Filtering midi messages before they reach LS



If you get a message : Engine: WARNING, CONFIG_EG_MIN_RELEASE_TIME too big for current audio fragment size & sampling rate! May lead to click sounds if voice stealing chimes in! and eventually a crash after that, then it's time to reconsider latency for your souncard. On a M-audio delta44 everything working at 5,8 ms.

External Ressources