LinuxSampler and Sonar 6

[jfcomposer]

Hi all,

I will apologize in advance for the length of this post.

I have finally decided to ditch Gigastudio 3 after several years of (mostly) faithful service. The search for a workable alternative is on, and I must say that I love the LinuxSampler interface and what it's capable of. I downloaded the Kontakt demo and had trouble getting into it, but with LinuxSampler I dove right in. Kudos!

Now for the issues

1. Main Issue: I've created an Orchestra, a MIDI device, and an audio device as the documentation says (standalone mode). I'm able to load a .gig instrument and hear it play with the keyboard within JSampler. The issue arises when I try to get playback from Sonar 6. I've set things up all sorts of ways and I just can't get any of the notes to sound. I'm currently running with a Sound Blaster Audigy 2 and an M-Audio Audiophile 2496. I've tried both as the MIDI inputs for JSampler and the Audio out (setting Sonar accordingly), and can't get anything. What am I missing here?

2. The documentation suggests that an "arbitrary" number of ports can be specified within JSampler for MIDI input. As far as I can tell, though, the number of ports is restricted by how many ports your soundcard tells Windows it has... in the case of the Audigy, it's 2, with the Audiophile, it's 1. With 16 channels each, that limits me to a maximum number of 48 discreet channels, as opposed to 128 with Gigastudio's "virtual ports" that it creates (and are selectable within Sonar). Am I correct in this conclusion or is there some trick to having an unlimited number of channels? This is important as I would routinely use 60+ channels for orchestral scores.

3. When trying to load LinuxSampler as a VST within Sonar, I get "A fatal error has occurred" with Exception code c0000005, in reference to LinuxSampler32.dll. I've tried several suggestions involving copying a couple dll's into the same folder as the Sonar exe, but haven't had any luck. Has anyone else experienced this or have any suggestions?

Thanks for any help you all may be able to provide. I really really want this to work and to get away from Gigastudio once and for all.

Jonathan

[ccherrett]

This is not the answer you are looking for but I thought I would let you know. A few of us have been developing a professional orchestral pipeline in Linux. You can follow the progress at http://www.openoctave.org. As for Window I know nothing. In Linux I have over 900 midi and audio ports. So at least in Linux I can have as many ports as I need.

[jfcomposer]

Thanks for the suggestion, but upon reading more about it I don't think it would fit into my current workflow. I've settled on Sonar for my sequencing needs and I'm pretty happy with it. To be honest, I would love to switch all this up to Linux, but I have yet to find a Linux equivalent of Sonar that has a focus on the staff view and video import. Also, most Linux apps do either audio or midi, and rarely both with any elegance. Even OpenOctave says on its website that notation (the part I'm most interested in) is not a primary goal of the project. I must admit that my workflow is probably different from most orchestra-oriented composers as I don't use the MIDI keyboard for real-time entry, so the piano roll view is much less comfortable than the staff view. Switching my sampler around is traumatic enough, but changing my sequencer as well is not something I'm particularly open to at this point. I just want to find a sampler that can load my .gig files and play them reasonably well (low latency, decent polyphony, etc.), not crash, and has at least a few of the nice features of Gigastudio, like GigaPulse and instrument layering.

I will certainly keep OpenOctave in mind for the future if I ever see a need for it. Thanks for the info!

[ccherrett]

I just want to find a sampler that can load my .gig files and play them reasonably well (low latency, decent polyphony, etc.), not crash, and has at least a few of the nice features of Gigastudio, like GigaPulse and instrument layering.

Sounds like the Open Octave Pipeline to me

You can use denemo for notation. My wife is the musician and plays in with a yamaha keyboard, so I am not dealing with notation much. oomidi has no notation anymore, we rely on other apps to do that.

As for having separate apps to do what is done in Windows or Mac with one, I would say is the power of the modular nature of linux audio. I have written a startup script that connects all the apps together at start up and restarts if an crashes happen. I will be releasing it soon.

I am using a custom jconv setup tied together by jack through ardour. It is very very nice with control for Early Reflections and tail per instrument, then controlled by a master reverb tail and ER bus. So one instance of the convolution engine running at 5% CPU for a full orchestral score and sounding fantastic. No sure how much better it gets

[ccherrett]

One last thing. The Open Octave Project is not about one application. We forked rosegarden 1.7.3 so we could get more control over professional midi editing, however we relay on many apps to get the job done.

Currently it looks like this for me:

Jack -> Linuxsampler -> OOMidi - > A2JMIDID -> ARDOUR -> JCONV -> ARDOUR -> System Out

So the whole orchestra running real time and access to every articulation without ever having to load anything up again.

Anyhow

[jfcomposer]

Jack -> Linuxsampler -> OOMidi - > A2JMIDID -> ARDOUR -> JCONV -> ARDOUR -> System Out

This is the sort of thing that makes me leery of dedicating my music setup to Linux. Even if I can get it to work, there are so many pieces in the puzzle that the solution becomes very complicated. Expressed the same way, my current setup is:

Sonar -> Gigastudio -> System Out

So essentially instead of replacing one piece of the puzzle with one piece (two at most), we're talking about replacing both pieces of the puzzle with 3 pieces each. Some people like the modularity of linux, that all applications do one thing and do it well. There is value to that philosophy, for sure, but when taken to extremes it becomes cumbersome. For instance, the Gimp can do many different things within the context of editing a photo, but let's say that you had to use one app for drawing a line, another for adjusting brightness/contrast, another for adding text, and yet another for raster touch-up. Highly modularized, but absurd all the same. I guess myself (and most of the rest of the world) are used to applications that lean more toward a full package, like Sonar which can handle MIDI, Audio, Video, and everything in between. The Gimp is a full package, as is OpenOffice. Why should linux music production be any different?

Please don't take this to mean I am unappreciative of your suggestions, because I certainly am appreciative of anyone who can shed light on how I might get things to work. So I guess it eventually comes down to how badly I want to get rid of my one remaining Windoze license and learn to make all of this work in Linux. Nay, not just work, but work effectively, elegantly, and productively, with few headaches. Because when the inspiration hits, the last thing I want to be doing is fighting with my setup.

Thanks again, and if anyone out there has suggestions on my original issue with LinuxSampler and Sonar on Windows, I'd love to hear it.

[ccherrett]

Sonar -> Gigastudio -> System Out cannot produce the work flow that is possible in the Open Octave Pipeline.

But the proof is in the pudding and I will have a video out soon to show what we have come out with.

Until then I hope you get your setup working. Linuxsampler is truly an awesome piece of software!

[ggoodesa]

Hi Jonathan,
I use LinuxSampler in Windows, so perhaps I can help a bit. I'm not using Sonar 6 (but I have friend who has it so I will see if I can get him to install LinuxSampler and see if we can get it working for him). For MIDI ports I use the virtual MIDI port driver from the creator of MIDI-OX, called MIDI Yoke, this can be configured to provide 16 virtual MIDI ports (each with 16 channels) giving a total of 256 midi channels to play with. You can download MIDI Yoke from http://www.midiox.com/ (just click on the MIDI Yoke link when you get there). When running LinuxSampler in Standalone, MIDI Yoke can provide the connections between Sonar and LinuxSampler. MIDI Yoke is a 32bit MIDI driver, so if you're running Windows 64bit you'll have purchase the LoopBe virtual MIDI port provider.
In regard to getting Audio out of LinuxSampler, do you get sound when using the keyboard in jSampler Fantasia?
Hope MIDI Yoke helps,
GrahamG

[jfcomposer]

GrahamG - Thanks for the info, I will certainly try MIDI Yoke tonight when I get home. It sounds as though this could be the missing link I'm looking for right now. Now if only Sonar would work in Wine... To answer your question: yes, I do get audio when clicking the keyboard in Fantasia. I'm pretty certain that once Sonar and LinuxSampler get coupled up right, things will work well.

In the mean time, there are only two other issues that I can see with making the complete switch: instrument layering and Gigapulse.

I'll post back with the results of my attempts this evening.

[ggoodesa]

Hi again Jonathan,
In regard to convolution reverb in Windows.. give Freeverb3 Impulser vst a try... it's opensource and can use most IR wav files that are downloadable on the net.

'Instrument layering' I don't know what that is (I'm not a Gigastudio user) so can't help there.

GrahamG