[Linuxsampler-devel] [linuxsampler-devel] SmartMidi Legato Instruments engine support

[Christian Schoenebeck]

I'd consider the 'correct' behaviour of change_note/set_event_par to be to
change the incoming midi note data itself, as opposed to the voice data,
but the use of the two in tandem could give interesting results.
Could I suggest an extension of change_note, an optional argument to change
the voice data itself and/or the incoming midi data?
This might be silly feature creep though.

I think I already addressed that today (as I was already at it when I added
change_velo() and change_note() today). After you updated LS to latest SVN it
should apply the note/velocity changes to the MIDI event as well as to the
voices, so it should behave as expected by you. Which is a substantial
behavior difference of course, since other samples would be triggered, mickey
mouse effect, etc.

I am actually not sure what the precise behavior of Kontakt is on that issue.
I guess it also modifies the MIDI event directly, but not sure.

Fact is Kontakt has two separate levels of event handling by scripts. One
level are the event handlers which we also have right now in LS. Then they
also have another raw MIDI event handler level called "Multi Script", which is
modifying the raw MIDI data itself before it is actually passed to that 2nd
level we have. That raw MIDI handling level of KSP has exactly one event
handler:

on midi_in
...
end on

It allows you to modify MIDI events, drop MIDI events and create completely
new MIDI events on your own by script.

ATM we don't have that event handler (and thus this first level of raw MIDI
processing) yet. Mainly because so far I saw no huge benefit of having it. But
who knows, maybe I am also missing out a killer use case for that layer.

CU
Christian

[Andrew C]

Just updated to the latest and it works (ignoring event_id ignores input).
Just having a logic problem in my script (My issue, not the sampler)

When I put in the full parameters for play_note and tell it to keep the
same note_off value for the original note, in 'case 0', the $event_note is
played correctly, but in 'case 1', the $event_note and $event_note + 5 get
played the same time. I thought I had told ignore_event to ignore the
initial note?

If I ignore event_id, the whole instrument because silent (well, this is
the correct behaviour of course :D )


--

On a related note, it seems that 'change_note' in ksp could achieve the
desired effect of transposing incoming notes:

"The values of $EVENT_NOTE and $EVENT_VELOCITY can be changed by using the
following functions:

change_note(<ID-number>,<new-note-number>)

change the note value of a specific note event

change_velo(<ID-number>,<new-velocity-number>)
change the velocity value of a specific note event"

Andrew.

on init
declare polyphonic $alt
end on

on note
ignore_event($EVENT_NOTE)

message("Alt is: " & $alt)

select $alt

case 0
play_note($EVENT_NOTE, $EVENT_VELOCITY, 0, -1)
message("Root note")
$alt := 1
case 1
play_note($EVENT_NOTE+5, $EVENT_VELOCITY, 0, -1)
message("Transposed")
$alt := 0
end select

end on

On Wed, May 24, 2017 at 9:12 PM, Christian Schoenebeck <

One question, how does "ignore_event" work? I RTFM and it says it can be
used to ignore MIDI note-on and note-off commands before they're

triggered.

I would imagine this *should* make the current note in $EVENT_ID not be
played, but when I click on the piano in gigedit/jsampler, the note plays
anyway.
note_off($EVENT_ID) in a "on note" block causes linuxsampler to segfault
too. :\

I haven't tried your script yet. However I just fixed a bunch of event
scheduling bugs which might potentially be related to what you
encountered. So
please update LS to latest SVN and try again. If the script still does not
work as expected, let me know!
CU
Christian
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Linuxsampler-devel mailing list
https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel

[Christian Schoenebeck]

Post by Andrew C
Just updated to the latest and it works (ignoring event_id ignores input).
Just having a logic problem in my script (My issue, not the sampler)
When I put in the full parameters for play_note and tell it to keep the
same note_off value for the original note, in 'case 0', the $event_note is
played correctly, but in 'case 1', the $event_note and $event_note + 5 get
played the same time. I thought I had told ignore_event to ignore the
initial note?

[snip]

Post by Andrew C
on init
declare polyphonic $alt
end on
on note
ignore_event($EVENT_NOTE)
message("Alt is: " & $alt)
select $alt
case 0
play_note($EVENT_NOTE, $EVENT_VELOCITY, 0, -1)
message("Root note")
$alt := 1
case 1
play_note($EVENT_NOTE+5, $EVENT_VELOCITY, 0, -1)
message("Transposed")
$alt := 0
end select
end on

Your variable $alt is declared "polyphonic". That means each time your note
event handler is executed, it is using a completely separate instance of your
variable $alt (which is always initialized with 0 by the way). Hence $alt is
always 0 and thus your case 1 is never executed.

Just remove keyword "polyphonic" and it will probably behave as you intended
it to do.

http://doc.linuxsampler.org/Instrument_Scripts/NKSP_Language/#polyphonic_variables

The value of polyphonic variables is only preserved from a note handler to its
(own) release handler. So you can use polyphonic variables to "stick" your own
custom values to one specific note if your will.

Post by Andrew C
On a related note, it seems that 'change_note' in ksp could achieve the
"The values of $EVENT_NOTE and $EVENT_VELOCITY can be changed by using the
change_note(<ID-number>,<new-note-number>)
change the note value of a specific note event
change_velo(<ID-number>,<new-velocity-number>)
change the velocity value of a specific note event"

Correct, that's what you rather wanted to do in your example instead. Right
now your can use set_event_par() instead:

http://doc.linuxsampler.org/Instrument_Scripts/NKSP_Language/Reference/set_event_par_function/

I will certainly also add change_vol() and change_note() just for sake of
completeness. The behavior is absolutely identical though.

ATM you can even do that:

$EVENT_NOTE := 40
$EVENT_VELOCITY := 127

But I am not sure yet whether that will change in future, maybe those will
become read-only variables one day, so writing to those built-in variables is
currently not a documented behavior, and probably you are not allowed to do
that with KSP at all.

CU
Christian

[Andrew C]

Hope this didn't come across as presumptous/demanding.. Just slightly
excited at the prospect of in-engine legato being a reality in LinuxSampler
from a design POV. Just seems really cool if it could be done and work.

So I have the Bb Clarinet long notes file mostly working out of the box.
Normal sustain/staccato programs work fine.
The X-fade instruments work fine.
The basic No Vibrato/Vibrato keyswitch instruments work fine.
*The sustain/attack mod wheel instruments need the attack attenuation
controller set to Modwheel 1 in gigedit for the attack sample intensity via
modwheel to work properly.
*Glissando/Legato instruments (using the imidi legato rules and smartmidi
The lowest note in the instrument's range is silent and just 'clicks' when
activated.
The adjacent C# appears to be programmed exactly the same way but sound is
heard(see below).
The notes above C4 or so are 'clicky' too for no apparent reason.
When playing the notes that do work, only the initial attack and the
sustain sample is heard, moving in a legato manner up/down the keyboard
does not work.
I'm not sure if the developers want to troubleshoot this for a single
unique giga file?
AFAICS, the imidi stuff is deprecated basically in favour of the kontakt
scripting (which'd make a lot more sense, but would require reprogramming
the entire legato instrument from the sample level up), but there is not
(yet) a working proof of concept of these rules working in a gig file under
LinuxSampler.
Andrew.

[Andrew C]

Post by Christian Schoenebeck
After you updated LS to latest SVN it
should apply the note/velocity changes to the MIDI event as well as to the
voices, so it should behave as expected by you.

Well, I updated to the latest svn, but the MIDI data itself doesn't change
with change_note unfortunately, only the voice data. i.e chipmunk effect
occurs if I transpose C3 up to C4 using change_note, rather than triggering
the C4. :S

Post by Christian Schoenebeck
ATM we don't have that event handler (and thus this first level of raw MIDI
processing) yet. Mainly because so far I saw no huge benefit of having it. But
who knows, maybe I am also missing out a killer use case for that layer.

I can dig that. From what I read of other Kontakt scripts, the multiscript
stuff is mainly used for channel switching tricks, etc.



On Thu, May 25, 2017 at 7:46 PM, Christian Schoenebeck <

Post by Christian Schoenebeck

I'd consider the 'correct' behaviour of change_note/set_event_par to be

to

change the incoming midi note data itself, as opposed to the voice data,
but the use of the two in tandem could give interesting results.
Could I suggest an extension of change_note, an optional argument to

change

the voice data itself and/or the incoming midi data?
This might be silly feature creep though.

I think I already addressed that today (as I was already at it when I added
change_velo() and change_note() today). After you updated LS to latest SVN it
should apply the note/velocity changes to the MIDI event as well as to the
voices, so it should behave as expected by you. Which is a substantial
behavior difference of course, since other samples would be triggered, mickey
mouse effect, etc.
I am actually not sure what the precise behavior of Kontakt is on that issue.
I guess it also modifies the MIDI event directly, but not sure.
Fact is Kontakt has two separate levels of event handling by scripts. One
level are the event handlers which we also have right now in LS. Then they
also have another raw MIDI event handler level called "Multi Script", which is
modifying the raw MIDI data itself before it is actually passed to that 2nd
level we have. That raw MIDI handling level of KSP has exactly one event
on midi_in
...
end on
It allows you to modify MIDI events, drop MIDI events and create completely
new MIDI events on your own by script.
ATM we don't have that event handler (and thus this first level of raw MIDI
processing) yet. Mainly because so far I saw no huge benefit of having it. But
who knows, maybe I am also missing out a killer use case for that layer.
CU
Christian
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Linuxsampler-devel mailing list
https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel

[Andrew C]

Thanks alot for fixing this, Christian!
My script works like a charm now - Already I can see interesting ways to
expand it such as subtly varying the tuning of each note played for more
humanisation, etc.

I'd like to see if I can't code an approximation of a 'poorman's' legato
function in NKSP that works by activating the sustain pedal when
overlapping notes are detected and turning the instrument into a
'monophonic' voice. I can probably do some cool stuff with, as you said,
%KEY_DOWN[] and $EVENT_NOTE.

I'm gonna see what I can come up with over the weekend..

Andrew.



On Thu, May 25, 2017 at 11:07 PM, Christian Schoenebeck <

Post by Andrew C

Post by Christian Schoenebeck
After you updated LS to latest SVN it
should apply the note/velocity changes to the MIDI event as well as to

the

Post by Andrew C

Post by Christian Schoenebeck
voices, so it should behave as expected by you.

Well, I updated to the latest svn, but the MIDI data itself doesn't

change

Post by Andrew C
with change_note unfortunately, only the voice data. i.e chipmunk effect
occurs if I transpose C3 up to C4 using change_note, rather than

triggering

Post by Andrew C
the C4. :S

Right, there was a bit more involved, which I hopefully just fixed. At
least
* change_note() should now reselect the correct region and accordingly the
correct sample (no more mickey mouse).
and
* change_velo() should now also reselect the correct dimension region (if
you
have a "velocity" dimension in giga, and whatever the term is for sfz and
sf2).
It is still not changing the actual MIDI event though BTW, which in
practice
means certain MIDI processing tasks within the sampler like exclusive
groups
(a.k.a. key groups), key based round robin and the like are all not
affected
by those two script functions. For those particular use cases adding the
mentioned raw MIDI processing script level might make sense. As far as I
can
see it right now, changing the implementations of change_note() and
change_velo() to perform true MIDI event modification is probably not a
good
idea, since it would require all kinds of MIDI re-processing inside the
sampler engine, which would be inefficient.
CU
Christian
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Linuxsampler-devel mailing list
https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel

[Christian Schoenebeck]

Thanks for that update, Christian and for dealing with my inane feedback.

Ok, our repository server is old, but not *that* old. ;-) It's svn (Subversion
server).

libtool: compile: g++ -DHAVE_CONFIG_H -I. -I../.. -Wreturn-type
-ffast-math -g -O2 -pthread -MT parser.lo -MD -MP -MF .deps/parser.Tpo -c
parser.cpp -fPIC -DPIC -o .libs/parser.o
In file included from parser_shared.h:16:0,
tree.h: In member function ‘virtual void
tree.h:715:84: error: ‘memset’ was not declared in this scope
memset(&polyphonicIntMemory[0], 0, polyphonicIntMemory.size() *
sizeof(int));
^
Makefile:532: recipe for target 'parser.lo' failed
make[5]: *** [parser.lo] Error 1

I just fixed it. Should work now.

CU
Christian