EXS 24 Logic Sampler Users Group

Hi,
(back and all from southern France, wasn't too bad but not as nice as I was
thinking either, but anyways...)

As I can see, (almost?) all the content of the Hollowsun site has been
converted by now.
Nice group effort I think.
Now, who's gonna start a "patch of the week" thing, based on those samples?
I allready fooled around with some layers (of different instruments), making
intensive use of the EXSs mangling capabilities, but as I'm not a synth
kinda guy I'm having a hard time to get some proper results...

Allright, quite some time ago I made some patches, almost converting the EXS
into some sort of standard VA synth. The principle was to use a lot of
layers, all using keymapped samples of simple waveforms (sine, triangle, saw
etc.), then select the samples by the slider only (instead of using velocity
as a source). In short, I tried to replace the usual OSC(s) waveform
selection by layers using appropriate samples.
And while it didn't seem to be a bad idea at first, CPU useage is going all
through the wall for my 30+ layer patch.
Now, I seem to remember some update ReadMe, telling that each layer would
use up the full amount of played voices, no matter whether they are actually
sounding (read: their velocity zone is selected) or not. As I can't seem to
find this ReadMe anymore, could anybody confirm this behaviour? It'd almost
render my "turn the EXS into a VA synth" efforts useless as I wouldn't save
any CPU cycles (which has been my original intention) in case each and every
zone would be triggered, no matter if it fits the selected velocity zone or
not.
Well, maybe it's not too much of a great idea anyways, as you can never tell
from the EXS's UI which layer/waveform was selected, plus, the lack of being
able to combine different layers (representing OSCs) without "saving as" a
new patch would not allow constructing truly "fat" sounds. I'd like to know
though.

Regards,
Sascha

On a fine day, 21-06-2004, Sascha Franck wrote:

>The principle was to use a lot of
>layers, all using keymapped samples of simple waveforms (sine, triangle, saw
>etc.), then select the samples by the slider only (instead of using velocity
>as a source).
>
>[...]
>
>Now, I seem to remember some update ReadMe, telling that each layer would
>use up the full amount of played voices, no matter whether they are actually
>sounding (read: their velocity zone is selected) or not. As I can't seem to
>find this ReadMe anymore, could anybody confirm this behaviour?

Correct: if you do a sample select by e.g. mod-wheel (non-velocity), 
all layers are playing at all times.  When you do sample select by 
velocity (the default behaviour) only one layer is triggered instead. 
Layering samples and cycling through them with the mod-wheel can make 
for some pretty cool effect, but it does indeed eat up CPU quite fast.

-- 
Hendrik Jan Veenstra   h @ k n o w a r e . n l
Omega Art: http://www.omega-art.com/

--- In exs-users@yahoogroups.com, "Sascha Franck" <S.Franck@g...> wrote:
> Hi,
> (back and all from southern France, wasn't too bad but not as nice as I was
> thinking either, but anyways...)
> 
> As I can see, (almost?) all the content of the Hollowsun site has been
> converted by now.
> Nice group effort I think.
> Now, who's gonna start a "patch of the week" thing, based on those samples?
> I allready fooled around with some layers (of different instruments), making
> intensive use of the EXSs mangling capabilities, but as I'm not a synth
> kinda guy I'm having a hard time to get some proper results...
> 
> Allright, quite some time ago I made some patches, almost converting the EXS
> into some sort of standard VA synth. The principle was to use a lot of
> layers, all using keymapped samples of simple waveforms (sine, triangle, saw
> etc.), then select the samples by the slider only (instead of using velocity
> as a source). In short, I tried to replace the usual OSC(s) waveform
> selection by layers using appropriate samples.
> And while it didn't seem to be a bad idea at first, CPU useage is going all
> through the wall for my 30+ layer patch.
> Now, I seem to remember some update ReadMe, telling that each layer would
> use up the full amount of played voices, no matter whether they are actually
> sounding (read: their velocity zone is selected) or not. As I can't seem to
> find this ReadMe anymore, could anybody confirm this behaviour? 

I find that looped single cycle waveforms use up a lot of CPU cycles vs. nonlooped 
samples. I created a patch with a single cycle sine wave and playing chords sucked 
the life out of my 867Mhz G4. You might try rendering each waveform to be several 
seconds long, and therefore the samples won't loop as much or at all.
-Matt Miller
http://www.millertone.com


It'd almost

> render my "turn the EXS into a VA synth" efforts useless as I wouldn't save
> any CPU cycles (which has been my original intention) in case each and every
> zone would be triggered, no matter if it fits the selected velocity zone or
> not.
> Well, maybe it's not too much of a great idea anyways, as you can never tell
> from the EXS's UI which layer/waveform was selected, plus, the lack of being
> able to combine different layers (representing OSCs) without "saving as" a
> new patch would not allow constructing truly "fat" sounds. I'd like to know
> though.
> 
> Regards,
> Sascha

Sascha Franck wrote: 

>
> Allright, quite some time ago I made some patches, almost converting the EXS
> into some sort of standard VA synth. The principle was to use a lot of
> layers, all using keymapped samples of simple waveforms (sine, triangle, saw
> etc.), then select the samples by the slider only (instead of using velocity
> as a source). In short, I tried to replace the usual OSC(s) waveform
> selection by layers using appropriate samples.


Apart from the CPU consumption problems you mention,  the EXS24 produces a lot
of aliasing when it is pitch shifting. So, unless you have one sample per
semitone at any velocity level (ie no pitch shifting),  it is going to end up
sounding worse than a lot of your VA source material. 

More recent VA synths use anti-aliasing techniques - eg Pentagon 1 and Z3ta+
have band limited oscillators - they sound much clearer than the early VA
synths like the Model E and the original Pro 5 because of the low levels of
aliasing noise.

I made a stack of synth sample based instruments a few years ago for the EXS24
using all the hardware samples I could get my hands on -  I never use them
these days because they are just too muddy sounding due to the accumulation of
aliasing noise. I would like to convert them all to SFZ format someday and hear
what they sound like without the aliasing. 

One solution would be for the EXS24 to have more levels of interpolation
quality like SFZ - or even a high quality render mode for bouncing.

Regards,
Murray