EXS 24 Logic Sampler Users Group

Anyone have any suggestions for improving EXS performance (or modifying 
it's voice stealing method) when dealing with voice hungry instruments like 
the Bosendorfer Imperial Grand?  I seem to be getting a lot of voice dropping 
on the Bosendorfer when using the pedal heavily, those 64 voices go fast it 
seems.  I've read some of the past posts on creative note stretching in the 
matrix after the fact, but  that seems like a time consuming work around.  I love 
the Bosendorfer, & EXS, but it seems it's voice stealing could perhaps be 
better so the drop outs are not so noticable in pedal heavy playing.  Any 
suggestions would be appreciated.

______
LAP 6 OSX
G4 1.25

On a fine day, 21-03-2003, lawbreak_a wrote:

>Anyone have any suggestions for improving EXS performance (or modifying
>it's voice stealing method) when dealing with voice hungry instruments like
>the Bosendorfer Imperial Grand?  I seem to be getting a lot of voice dropping
>on the Bosendorfer when using the pedal heavily, those 64 voices go fast it
>seems.  I've read some of the past posts on creative note stretching in the
>matrix after the fact, but  that seems like a time consuming work 
>around.  I love the Bosendorfer, & EXS, but it seems it's voice 
>stealing could perhaps be
>better so the drop outs are not so noticable in pedal heavy playing.  Any
>suggestions would be appreciated.

Sounds like you simply want more than 64 voices, right?  I never 
tried the following myself, but imo it should work.  Basic idea: 
connect a transformer to 2 EXS instances, and have the transformer 
send all notes below e.g. C3 to the first instance and the other 
notes to the second instance (assuming your entire part isn't played 
with 2 hands below C3 :-).  At the same time, you should make sure 
that pedal messages (CC64) reach both EXS's.

Details: create a Monitor object (or any other 'neutral' object) in 
the Audio Mixer layer of the environment, and check its icon in its 
parameter pane.
Create 2 transformers, and cable the Monitor into both transformers 
(which are thus in parallel).  Open the 1st transformer 
(double-click) and set the top popup "Apply Operation and filter 
non-matching events".  Set Conditions thus:
Status = Control
Cha = <whatever, or leave alone>
-1- = 64
-2- leave alone.

This transformer will now only pass CC64 (sustain) messages.  Cable 
this transformer into the 1st EXS channel strip only.

Open the 2nd transformer, and set its top popup to "Condition 
Splitter".  Set Conditions as follows:
Status = Note
Cha = <whatever>
-1- = '<='  C3 (or wherever you want the split to occur)
-2-  leave alone

Cable the *1st* outlet of the transformer into the 1st EXS instance 
(which now receives notes below C3, plus CC64 through the 1st 
transformer).  Cable the *2nd* outlet to the 2nd EXS (which receives 
everything which is not "note below C3", which thus includes CC64 
messages).

So now you have:
Monitor into Transf.1 & 2.  Transf.1 into EXS1.  Transf.2 into EXS1 & 2.

Insert an I/O plugin in both EXS's, and bypass them -- this is the 
trick which allows realtime playing of multiple VSTi's.

Assign an Arrange track to the Monitor object, and select that track 
(or put your midi sequences on that track).  Oh, you might want to 
set the output of both EXS channel strips to a bus, and use the bus 
for effects inserts, volume automation and such.

As I said: I never tried this myself, but as far as I can tell, it 
should work.  Would be curious to hear if it indeed works (if you 
decide to give it a try, that is).

-- 
Hendrik Jan Veenstra  <h@...>
Omega Art: http://www.omega-art.com

Wow, I will try this out this weekend & get back with details.  this 
would completely solve my problems!  Anyone know if theres ever 
going to be a 128 voice EXS?
Thanks!

--- In exs-users@yahoogroups.com, Hendrik Jan Veenstra <h@k...> 
wrote:
> On a fine day, 21-03-2003, lawbreak_a wrote:
> 
> >Anyone have any suggestions for improving EXS performance (or 
modifying
> >it's voice stealing method) when dealing with voice hungry 
instruments like
> >the Bosendorfer Imperial Grand?  I seem to be getting a lot of 
voice dropping
> >on the Bosendorfer when using the pedal heavily, those 64 voices 
go fast it
> >seems.  I've read some of the past posts on creative note 
stretching in the
> >matrix after the fact, but  that seems like a time consuming work 
> >around.  I love the Bosendorfer, & EXS, but it seems it's voice 
> >stealing could perhaps be
> >better so the drop outs are not so noticable in pedal heavy 
playing.  Any
> >suggestions would be appreciated.
> 
> Sounds like you simply want more than 64 voices, right?  I never 
> tried the following myself, but imo it should work.  Basic idea: 
> connect a transformer to 2 EXS instances, and have the transformer 
> send all notes below e.g. C3 to the first instance and the other 
> notes to the second instance (assuming your entire part isn't 
played 
> with 2 hands below C3 :-).  At the same time, you should make sure 
> that pedal messages (CC64) reach both EXS's.
> 
> Details: create a Monitor object (or any other 'neutral' object) 
in 
> the Audio Mixer layer of the environment, and check its icon in 
its 
> parameter pane.
> Create 2 transformers, and cable the Monitor into both 
transformers 
> (which are thus in parallel).  Open the 1st transformer 
> (double-click) and set the top popup "Apply Operation and filter 
> non-matching events".  Set Conditions thus:
> Status = Control
> Cha = <whatever, or leave alone>
> -1- = 64
> -2- leave alone.
> 
> This transformer will now only pass CC64 (sustain) messages.  
Cable 
> this transformer into the 1st EXS channel strip only.
> 
> Open the 2nd transformer, and set its top popup to "Condition 
> Splitter".  Set Conditions as follows:
> Status = Note
> Cha = <whatever>
> -1- = '<='  C3 (or wherever you want the split to occur)
> -2-  leave alone
> 
> Cable the *1st* outlet of the transformer into the 1st EXS 
instance 
> (which now receives notes below C3, plus CC64 through the 1st 
> transformer).  Cable the *2nd* outlet to the 2nd EXS (which 
receives 
> everything which is not "note below C3", which thus includes CC64 
> messages).
> 
> So now you have:
> Monitor into Transf.1 & 2.  Transf.1 into EXS1.  Transf.2 into 
EXS1 & 2.
> 
> Insert an I/O plugin in both EXS's, and bypass them -- this is the 
> trick which allows realtime playing of multiple VSTi's.
> 
> Assign an Arrange track to the Monitor object, and select that 
track 
> (or put your midi sequences on that track).  Oh, you might want to 
> set the output of both EXS channel strips to a bus, and use the 
bus 

> for effects inserts, volume automation and such.
> 
> As I said: I never tried this myself, but as far as I can tell, it 
> should work.  Would be curious to hear if it indeed works (if you 
> decide to give it a try, that is).
> 
> -- 
> Hendrik Jan Veenstra  <h@...>
> Omega Art: http://www.omega-art.com

I also wonder if you can split the track into two and have each one play 
half the notes and an instance of EXS on each.

Colin Miller

At 06:42 PM 3/21/2003 +0000, you wrote:

>Wow, I will try this out this weekend & get back with details.  this
>would completely solve my problems!  Anyone know if theres ever
>going to be a 128 voice EXS?
>Thanks!

> Sounds like you simply want more than 64 voices, right?  I never 
> tried the following myself, but imo it should work.  Basic idea: 
> connect a transformer to 2 EXS instances, and have the transformer 
> send all notes below e.g. C3 to the first instance and the other 
> notes to the second instance (assuming your entire part isn't 
played 
> with 2 hands below C3 :-).  At the same time, you should make sure 
> that pedal messages (CC64) reach both EXS's.

Hendrik, first idea I thought of is a switcher that switches every 
other note to one of the two EXS 24's. This would divide up the work 
more evenly but I have not a clue as to how to do this.

Bob Vandiver

Yes I've tried this & it works fine, but for live situations (which 
would be nice) I would like to find a fix as well.  Hopefully the 
environment splitter suggestion earlier will help.

--- In exs-users@yahoogroups.com, Colin Miller <snoopy@m...> wrote:
> I also wonder if you can split the track into two and have each 
one play 
> half the notes and an instance of EXS on each.
> 
> Colin Miller
> 
> At 06:42 PM 3/21/2003 +0000, you wrote:
> >Wow, I will try this out this weekend & get back with details.  
this

> >would completely solve my problems!  Anyone know if theres ever
> >going to be a 128 voice EXS?
> >Thanks!

On a fine day, 21-03-2003, bob_vandiver wrote:

>  > Sounds like you simply want more than 64 voices, right?  I never
>>  tried the following myself, but imo it should work.  Basic idea:
>>  connect a transformer to 2 EXS instances, and have the transformer
>>  send all notes below e.g. C3 to the first instance and the other
>>  notes to the second instance (assuming your entire part isn't
>played
>>  with 2 hands below C3 :-).  At the same time, you should make sure
>>  that pedal messages (CC64) reach both EXS's.
>
>Hendrik, first idea I thought of is a switcher that switches every
>other note to one of the two EXS 24's. This would divide up the work
>more evenly but I have not a clue as to how to do this.

Ah, yes, clever idea...  Well, you could use a transformer set to 
"alternating split", which does exactly what the name says: send out 
events to outlets 1 and 2 in turn.  The problem with that though is 
that you could inadvertently  send a note-on to one outlet and the 
note-off to another (ditto with e.g,. sustain pedal on/off messages). 
The easiest fix would then be to first split the incoming midi stream 
into 'note ons' and 'everything else'.  Send 'everything else' (i.e. 
including note-offs) to both EXS's, and send the note-ons to an 
"Alternating Split" transformer.  A bit tricky, since it means you 
have to convert notes to CC's first, or else the environment won't 
let you spliut note-ons and note-offs, etc.  Also, both EXS's will 
recieve _all_ note-offs, even for notes they didn't receive a note-on 
for, which doesn't hurt but is a bit messy of course.

-- 
Hendrik Jan Veenstra  <h@...>
Omega Art: http://www.omega-art.com

On a fine day, 21-03-2003, lawbreak_a wrote:

>--- In exs-users@yahoogroups.com, Hendrik Jan Veenstra <h@k...>
>wrote:
>  > Sounds like you simply want more than 64 voices, right?  I never
>  > tried the following myself, but imo it should work.
>[...]
>
>Wow, I will try this out this weekend & get back with details.  this
>would completely solve my problems!  Anyone know if theres ever
>going to be a 128 voice EXS?
>Thanks!

Just tried it myself -- one Monitor object and 2 transformers, less 
than 5 minutes to set up, and it works beautifully.  Note that Logic 
has to be playing for this to work though -- while stopped only the 
1st EXS will be playable.

This was also the 1st time I (finally) tried the "bypassed I/O-plugin 
to use multiple VSTi's" trick.  It does work, but the latency is 
quite noticeable imo.  I had to set the I/O buffer to minimum to get 
sort-of acceptable playability.  The 1st EXS doesn't suffer from 
that, but the 2nd does.  So you might want to fiddle a bit with your 
hardware settings to get the best result.  Don't know if in your case 
it's good enough for live-use, but I guess this (and variations on 
the theme, like the one Bob Vandiver suggested) is the only way to 
circumvent the 64-voice limitation.

One more thought... I remember conflicting reports on the 
latency-subject when the I/O workaround was 1st discussed on the LUG. 
Some reporting that the 1st instance didn't need the I/O plug and 
others claiming that it did -- all having to do with getting at least 
the _same_ amount of latency on both instances...  No time to check 
this out myself at the moment (already quit Logic, didn't save the 
experiment, and a busy schedule ahead), but that's another thing you 
might want to try: remove the I/O plugin from the 1st EXS and see if 
that makes a difference...

And if this split-keyboard-thing doesn't suffice, I wouldn't mind 
giving Bob's "alternating split" idea a try (not today though :). 
Just say so if you're interested.  If the split suffices, I wouldn't 
bother with more complex setups though.

-- 
Hendrik Jan Veenstra  <h@...>
Omega Art: http://www.omega-art.com

From: Hendrik Jan Veenstra <h@...>

>> Anyone have any suggestions for improving EXS performance (or modifying
>> it's voice stealing method) when dealing with voice hungry instruments
>> like the Bosendorfer Imperial Grand?  I seem to be getting a lot of voice
>> dropping on the Bosendorfer when using the pedal heavily, those 64 voices
>> go fast it seems..

> Sounds like you simply want more than 64 voices, right? .. Basic idea:
> connect a transformer to 2 EXS instances, and have the transformer send
> all notes below e.g. C3 to the first instance and the other notes to the
> second instance (assuming your entire part isn't played with 2 hands below
> C3 :-).  At the same time, you should make sure that pedal messages (CC64)
> reach both EXS's.

Hi Hendrik -

It might also be worth trying the following: Divide low and high notes into
2 groups _within_ the Instrument. Since one group apparently won't steal
notes from another, this might help to manage the voice stealing so it
isn't as noticeable musically.. For example, with piano the most noticeable
stealing I hear is long ringing bass notes being obviously stolen by
right-hand noodling.. splitting high/low notes by groups often stops this
for me; of course notes in the upper-register group are still being stolen
by the noodling, but not as noticeably as hanging bass notes cutting out,
often not noticeably at all..

I expect this won't extend the maximum polyphony of an EXS instance, and
when max polyphony is reached then assumedly any notes can be stolen, so if
playing style or layering of samples results in max polyphony being reached
regularly, then of course the dual-instance solution would be the way to
go.. However, the dual-group approach might manage the problem enough
(depending on musical style) to be an acceptable workaround sometimes,
avoiding the need, in those cases. to fudge with i/o plugs etc..  :-)

Cheers, Joe Albano
ROOFTOP PRODUCTIONS, NYC NY
Music: www.rooftopproductions.com
Consulting: www.rooftopproductions.com/Consulting.html

You guys are all obviously environment gurus. I would tackle the problem 
from a different angle. What I would do is take the sequence in the arrange 
page, open it in the matrix editor, cut half of the notes, and paste them 
into another sequence. Then all you gotta do is create an audio instr. w/EXS 
and same instrument and put the second sequence on it. Voila! Stop thinking 
so hard! You can do every other note or all notes above (eg) C3 or whatever 
you like.

There was a story of a group of very intelligent scientists at a convention. 
A group of them went to lunch on the last day. When they sat down, they 
noticed that the salt shaker had pepper in it and the pepper shaker had salt 
in it. They proceeded to devise a plan to switch the contents of the shakers 
without spilling any and without using a third object. When the waitress 
came and asked what they were doing, they informed her of the situation and 
their plan to solve it. She then said "oh sorry," and proceeded to switch 
the tops!

_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online  
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963

At 02:26 PM 3/22/2003 -0500, you wrote:
>You guys are all obviously environment gurus. I would tackle the problem
>from a different angle. What I would do is take the sequence in the arrange
>page, open it in the matrix editor, cut half of the notes, and paste them
>into another sequence. Then all you gotta do is create an audio instr. w/EXS
>and same instrument and put the second sequence on it. Voila! Stop thinking
>so hard! You can do every other note or all notes above (eg) C3 or whatever
>you like.

Their reason was for live purposes, where one can't split the track in half 
(which was my suggestion earlier). So looks like there is a method for each 
situation. :-)

Colin Miller

On a fine day, 22-03-2003, Mark Jones wrote:

>You guys are all obviously environment gurus. I would tackle the problem
>from a different angle. What I would do is take the sequence in the arrange
>page, open it in the matrix editor, cut half of the notes, and paste them
>into another sequence. Then all you gotta do is create an audio instr. w/EXS
>and same instrument and put the second sequence on it. Voila! Stop thinking
>so hard! You can do every other note or all notes above (eg) C3 or whatever
>you like.

Sure, you're right -- if you have a pre-recorded track this would be 
the most obvious way to solve a polyphony problem.  However, it won't 
work for live usage, and the environment trick would.  That's why I 
suggested it.  But yes, just cutting the track in 2 parts would be 
the easiest way out for pre-recorded material.

>There was a story of a group of very intelligent scientists at a convention.
>A group of them went to lunch on the last day. When they sat down, they
>noticed that the salt shaker had pepper in it and the pepper shaker had salt
>in it. They proceeded to devise a plan to switch the contents of the shakers
>without spilling any and without using a third object. When the waitress
>came and asked what they were doing, they informed her of the situation and
>their plan to solve it. She then said "oh sorry," and proceeded to switch
>the tops!

LOL! :-)  One question though: _did_ these scientists figure out a 
way to do it or not? :-)

-- 
Hendrik Jan Veenstra  <h@...>
Omega Art: http://www.omega-art.com