Re: [Linuxsampler-devel] new to the list, PMI

[<be...@ga...>]

Hi Michiel,
thanks for joining the list,
your feedback will be very valuable because we know of your high quality
standards when producing sample libraries which means you will be valuable
for spotting playback imperfections (eg wrong velocity mappings, etc)
in LS and of course giving advices and ideas for new development directions.

About installing Linux, as said I'll write a small document on our website that
explains in short how to set Planet CCRMA  which add on packages
for Red Hat Linux that turn it in linux distribution optimized for music, alias
kernel optimized for low latency (better real time response), audio drivers for
the high end audio cards (ALSA drivers) installs
an easy to use package management that permits the one click install of
precompiled music apps and system libraries etc.


Michiel, your P4 1.8 is fully Ok for LS testing, I perform some tests
on much slower boxes (Celeron 400, VIA 1GHz) and the performance is quite
good there. Testing on slower boxes makes it easier to spot performance problems
and helps to make the engine very fast.
Your P4 is pretty high end and depending from your HD it allows easily
for 120-150 stereo voices (or hundred of mono voices).
So it is fully ok for testing.
Regarding hollywood composers, well at first it was a bit of a half joke between
us developers, in the sense we want to satisfy even their professional needs,
wich means if we can satisfy them we can satisfy anyone.
Of course before LS is ready for Hollywood it must provide perfect GIG playback
and network clustering with VSTi frontends. So it will still take some time.
As you said our main target is probably the average (relatively poor) windows
users which want to have a powerful sampler at his disposal.

Of course there windows users are using their own windows sequencers and
don't want do give them up easily since these apps under Linux are not
so mature like under Windows yet.
So a good compromise is to use a second box with LS with a VSti/AudioUnit
frontend which permits working with LS as it was a local application.

I mean: exclude those that pirate a soft sampler for windows/mac.
The cost of such a sampler is usually $300-$500. With that amount of money
you can buy a quite powerful machine these days and dedicate it entirely to
sampling.
So LS provides more much performance for the money, since if you are running
a windows softsampler on the same box that runs the sequencers, the performance
 (and possibility to load large sample setups) will not be as high as when
using two machines.

Running Linux and Windows in parallel on the same machine is possibile but
is fleasible only for applications that do not need real time response,
like Office applications etc.
But for LS it would be impossible to run it in a Linux-emulation window under
Windows (using
an application like VMWare wich allows running multiple OSes at the same time).
The real time response and disk streaming performance would simply suck due
to the added overhead of emulation layer.

About porting LS to other operating systems.
This would be ideal for getting a big marketshare quickly.
For example since OS X is based on UNIX and the LS codebase is quite portable
it would probably be quite easy to produce a native OS X version with
a VSTi or AudioUnit interface.
Probably we will do the port as soon as full GIG support in LS is completed
and as soon as the GUI is finished.
We use the Qt GUI programming library which permits to port the GUI code
to multiple OSes (Linux,Mac, Windows and others) by simple recompilation.

So don't worry, as soon as the first public LS version is a native
OS X version (which will provide high performance too since OS X is optimized
for audio) will
 be produced too.
This will bring new developers, new ideas and other useful stuff that
will benefit the entire LS project (thus Linux users too).

Now to Windows:
would a Windows port of LS be possible.
Definitively. The LS audio engine is small and portable because you mostly
need only to add a new audio output method and midi input method to make it work.
The GUI (when ready, Rui is working on it) is in Qt too so it can be compiled
on Windows without modification.
So while a local VSTi on Windows is possible it would be plagued by the same
problems that afflict other plugs: it's hard to get latency
down while maintain dropout-free operation, streaming performance is usually
 lower than on Linux etc.
Anyway at least for me (I'm speaking when LS on Linux will work well) I'm
not against a native Mac/Win port.
Of course our wish would be all users switching to Linux but on a short-term
basis this is simply not going to happen.
So any new user/developer of LS would benefit the project, regardless of
the operating system they use.
(just like OpenOffice profits from Linux, Solaris, Windows and Mac users).

Of course for a Mac/Windows port we would need the help of experienced
developers but I think once they see the potential of LS they will join us.
For example, ardour a prootols-like audio sequencer for Linux (
http://ardour.sourceforge.net ) is achieving professional grade level and
people are interested to have an OS X port. Work is underway and it will not
be that hard since ardour's code is quite portable.
Once many Mac users will use ardour it will benefit Linux ardour users too.

We need to set small goals, one after one.
For now the first goal is to get perfect GIG playback and an easy to use
 GUI working so that people can use LS in production enviroments, or
can be used on dedicated LS "expanders" wich can run PMI's and other libraries
out of the box, without needing configuration etc.

Then after this step many things can be done, like
an open/flexible sampling format to accomodate sample producers
special needs, sampling engines which tune themselves to get high
performance out of low cost hardware etc.

Regarding copyprotection: is it a task to solve and dongles are not
the ideal solution since the software will probably be cracked after
some time.
I mostly agree with what Bruce A.R. said on the NS forums that individual
watermarking is probably the most flexible solution.
At least is helps to identify who posted the sample CD on the net and
will deter people form posting copies of their stuff on P2P apps or websites.

Anyway let's worry about this stuff after the initial LS version is finished,
bfore that it does not make sense since no one is going to use our sampler
if it does not sound well :-)

Comments and questions welcome.

cheers,
Benno
http://www.linuxsampler.org

Scrive Michiel Post <mp...@xs...>:

> Hi LS group, 
> 
> I´m new to this group so let me introduce myself to you guys.
> 


-------------------------------------------------
This mail sent through http://www.gardena.net