linuxsampler-devel

[Linuxsampler-devel] poorman's round robin script?

[Andrew C <cou...@gm...>]

Hi all,

I'm looking into creating a poorman's round robin script for some short
note staccatos.
My plan is to write a script which will:

At note on, pick a random number between 0 and 4.
Play either the same note, 2 semitones up but repitched to the original
note's pitch, 1 semitone up but repitched, 2 semitones down but repitched
or 1 semitone down but repitched.

I looked at the NKSP reference manual and it seems like I can achieve the
first part with the Random() function.

I'm considering using change_pitch function to simply tell the future note
to pitch itself up or down and then use play_note to play the new note's
sample but repitched.

play_note does send a new midi note_on event, if I'm not mistaken?

Thanks,

Andrew.

Re: [Linuxsampler-devel] poorman's round robin script?

[Christian S. <sch...@li...>]

On Sonntag, 4. April 2021 22:24:09 CEST Andrew C wrote:
> Hi all,
> 
> I'm looking into creating a poorman's round robin script for some short
> note staccatos.

I assume you want to use sfz. In this case create a text file somewhere which 
you will fill with NKSP script code and link that script in your sfz file with 
the 'script' opcode:

	http://doc.linuxsampler.org/sfz/script_opcode/

The advantage with the gig format is that you get immediate syntax / error / 
warning feedbacks while typing the script in gigedit's script editor. So this 
is probably more convenient and faster, especially on first steps.

However alternatively you can also use 'ls_instr_script' from the command line 
to check your script for errors:

	cat /some/where/foo.txt | ls_instr_script sfz

Additional options are described by:

	ls_instr_script

> My plan is to write a script which will:
> 
> At note on, pick a random number between 0 and 4.

on init
  declare polyphonic $n
end init

on note
  $n := random(0, 4)
  message("Random number is " & $n)
  ...
end on

> Play either the same note, 2 semitones up but repitched to the original
> note's pitch, 1 semitone up but repitched, 2 semitones down but repitched
> or 1 semitone down but repitched.

  select $n
    case 0
      play_note($EVENT_NOTE + 2, $EVENT_VELOCITY)
    case 1
      ...
    case 2
      ...
    case 3
      ...
    case 4  
      ...
  end select

I don't get though what you mean with "repitched to the original note's 
pitch". Do you want to start it with a x semitones offset and let it slide to 
the original pitch within a certain duration?

> I looked at the NKSP reference manual and it seems like I can achieve the
> first part with the Random() function.
> 
> I'm considering using change_pitch function to simply tell the future note
> to pitch itself up or down and then use play_note to play the new note's
> sample but repitched.

The function is called change_tune() for some reason (KSP invention):

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

And you would probably want to use it in conjunction with change_tune_time() 
to control how long it will take to perform the tuning/pitch change.

> 
> play_note does send a new midi note_on event, if I'm not mistaken?

Yes.

This tour is probabably a good starting point:

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

Hover your mouse over keywords to see a short summary what they are doing. 
Click on them to get to the detailed description of the keyword.

> 
> Thanks,
> 
> Andrew.

CU
Christian

Re: [Linuxsampler-devel] poorman's round robin script?

[Andrew C <cou...@gm...>]

Hi Christian,

Thanks so much for the code snippets, they'll get me off to a good start!
I'll be doing this wholly in gig format, not sfz.

What I mean by repitching the note is not a slide, but rather play for
example "$EVENT_NOTE +1" (A C# sample) at the same pitch as C
($EVENT_NOTE)..
That way if I have a semitone sampled instrument, I could effectively get
extra repitched samples per note.

I think for this I can simply use change_tune to do an "instant" -100 cent
tune down so the C# sample sounds at the C pitch without too much
stretching?

Cheers.
Andrew.

On Tue, Apr 6, 2021 at 10:21 AM Christian Schoenebeck <
sch...@li...> wrote:

> On Sonntag, 4. April 2021 22:24:09 CEST Andrew C wrote:
> > Hi all,
> >
> > I'm looking into creating a poorman's round robin script for some short
> > note staccatos.
>
> I assume you want to use sfz. In this case create a text file somewhere
> which
> you will fill with NKSP script code and link that script in your sfz file
> with
> the 'script' opcode:
>
>         http://doc.linuxsampler.org/sfz/script_opcode/
>
> The advantage with the gig format is that you get immediate syntax / error
> /
> warning feedbacks while typing the script in gigedit's script editor. So
> this
> is probably more convenient and faster, especially on first steps.
>
> However alternatively you can also use 'ls_instr_script' from the command
> line
> to check your script for errors:
>
>         cat /some/where/foo.txt | ls_instr_script sfz
>
> Additional options are described by:
>
>         ls_instr_script
>
> > My plan is to write a script which will:
> >
> > At note on, pick a random number between 0 and 4.
>
> on init
>   declare polyphonic $n
> end init
>
> on note
>   $n := random(0, 4)
>   message("Random number is " & $n)
>   ...
> end on
>
> > Play either the same note, 2 semitones up but repitched to the original
> > note's pitch, 1 semitone up but repitched, 2 semitones down but repitched
> > or 1 semitone down but repitched.
>
>   select $n
>     case 0
>       play_note($EVENT_NOTE + 2, $EVENT_VELOCITY)
>     case 1
>       ...
>     case 2
>       ...
>     case 3
>       ...
>     case 4
>       ...
>   end select
>
> I don't get though what you mean with "repitched to the original note's
> pitch". Do you want to start it with a x semitones offset and let it slide
> to
> the original pitch within a certain duration?
>
> > I looked at the NKSP reference manual and it seems like I can achieve the
> > first part with the Random() function.
> >
> > I'm considering using change_pitch function to simply tell the future
> note
> > to pitch itself up or down and then use play_note to play the new note's
> > sample but repitched.
>
> The function is called change_tune() for some reason (KSP invention):
>
> http://doc.linuxsampler.org/Instrument_Scripts/NKSP_Language/Reference/
> change_tune_function/
> <http://doc.linuxsampler.org/Instrument_Scripts/NKSP_Language/Reference/change_tune_function/>
>
> And you would probably want to use it in conjunction with
> change_tune_time()
> to control how long it will take to perform the tuning/pitch change.
>
> >
> > play_note does send a new midi note_on event, if I'm not mistaken?
>
> Yes.
>
> This tour is probabably a good starting point:
>
> http://doc.linuxsampler.org/Instrument_Scripts/NKSP_Language/
>
> Hover your mouse over keywords to see a short summary what they are doing.
> Click on them to get to the detailed description of the keyword.
>
> >
> > Thanks,
> >
> > Andrew.
>
> CU
> Christian
>
>
>
>
> _______________________________________________
> Linuxsampler-devel mailing list
> Lin...@li...
> https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel
>

Re: [Linuxsampler-devel] poorman's round robin script?

[Christian S. <sch...@li...>]

On Dienstag, 6. April 2021 10:39:55 CEST Andrew C wrote:
> Hi Christian,
> 
> Thanks so much for the code snippets, they'll get me off to a good start!
> I'll be doing this wholly in gig format, not sfz.

Ok, then just hit Ctrl + S whenever you changed something in gigedit's script 
editor. That will cause the script to be reloaded by the sampler to make your 
changes audible.

> What I mean by repitching the note is not a slide, but rather play for
> example "$EVENT_NOTE +1" (A C# sample) at the same pitch as C
> ($EVENT_NOTE)..
> That way if I have a semitone sampled instrument, I could effectively get
> extra repitched samples per note.
> 
> I think for this I can simply use change_tune to do an "instant" -100 cent
> tune down so the C# sample sounds at the C pitch without too much
> stretching?

The trick here is that play_note() returns the note ID of that 
programmatically triggered new note. So you would use that note id and pass it 
to change_tune() like:

on init
  declare polyphonic $staccatoNote
end init

on note
  ...
  $staccatoNote := play_note($EVENT_NOTE + $n, $EVENT_VELOCITY)
  { let's say drop pitch by 30 cents immediately }
  change_tune($staccatoNote, -30c)
end note

Because obviously you just want to change the tuning of that sample; not of 
the original one. If you want to change the original one as well, let's say 
bring some randomness there as well:

  change_tune($EVENT_NOTE, random(-30C, +30C))

I guess you get the point.

CU
Christian

Re: [Linuxsampler-devel] poorman's round robin script?

[Andrew C <cou...@gm...>]

Hi Christian/List,

I've written and fully tested a working "neighbour notes" round robin
script with some rudimentary logic to avoid "dead" notes at either extreme
of the instrument.
The initial instrument range is for a cello, but can easily be edited for
any instrument range.

I've uploaded it here (RRCello. LScript):
https://drive.google.com/file/d/18c6iL-zR-tDYNBFyqUS01HnZipwbvPuJ/view?usp=sharing

If nothing else, it might serve as a good "real world" example of NKSP in
Linuxsampler.

Cheers,

Andrew.


On Tue, Apr 6, 2021 at 11:10 AM Christian Schoenebeck <
sch...@li...> wrote:

> On Dienstag, 6. April 2021 10:39:55 CEST Andrew C wrote:
> > Hi Christian,
> >
> > Thanks so much for the code snippets, they'll get me off to a good start!
> > I'll be doing this wholly in gig format, not sfz.
>
> Ok, then just hit Ctrl + S whenever you changed something in gigedit's
> script
> editor. That will cause the script to be reloaded by the sampler to make
> your
> changes audible.
>
> > What I mean by repitching the note is not a slide, but rather play for
> > example "$EVENT_NOTE +1" (A C# sample) at the same pitch as C
> > ($EVENT_NOTE)..
> > That way if I have a semitone sampled instrument, I could effectively get
> > extra repitched samples per note.
> >
> > I think for this I can simply use change_tune to do an "instant" -100
> cent
> > tune down so the C# sample sounds at the C pitch without too much
> > stretching?
>
> The trick here is that play_note() returns the note ID of that
> programmatically triggered new note. So you would use that note id and
> pass it
> to change_tune() like:
>
> on init
>   declare polyphonic $staccatoNote
> end init
>
> on note
>   ...
>   $staccatoNote := play_note($EVENT_NOTE + $n, $EVENT_VELOCITY)
>   { let's say drop pitch by 30 cents immediately }
>   change_tune($staccatoNote, -30c)
> end note
>
> Because obviously you just want to change the tuning of that sample; not
> of
> the original one. If you want to change the original one as well, let's
> say
> bring some randomness there as well:
>
>   change_tune($EVENT_NOTE, random(-30C, +30C))
>
> I guess you get the point.
>
> CU
> Christian
>
>
>
>
> _______________________________________________
> Linuxsampler-devel mailing list
> Lin...@li...
> https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel
>

Re: [Linuxsampler-devel] poorman's round robin script?

[Christian S. <sch...@li...>]

On Mittwoch, 7. April 2021 16:31:10 CEST you wrote:
> Hi Christian/List,
> 
> I've written and fully tested a working "neighbour notes" round robin
> script with some rudimentary logic to avoid "dead" notes at either extreme
> of the instrument.
> The initial instrument range is for a cello, but can easily be edited for
> any instrument range.
> 
> I've uploaded it here (RRCello. LScript):
> https://drive.google.com/file/d/18c6iL-zR-tDYNBFyqUS01HnZipwbvPuJ/view?usp=s
> haring
> 
> If nothing else, it might serve as a good "real world" example of NKSP in
> Linuxsampler.

Just some notes/tips from my side:

> on init
> {Round robin script for Cello. Version 0.1}
> {Written by Andrew Coughlan. Released under GPL v2 or later.}

Maybe a more detailed description what the script does would make sense. E.g.

"Assumes an instrument with one sample per key; it emulates round robin
samples by randomly picking another sample (from another region) and
pitching it up or down accordingly."

> 
>   declare polyphonic $n
>   declare polyphonic $RR
>   declare polyphonic $lokey := 36 {lowest playable note of this instrument}
>   declare polyphonic $hikey := 76 {highest playable note of this instrument}
> 
> end on

You could automate retrieval of $lokey and $hikey, at least for gig sounds: by
calling same_region($i, $i) in a loop inside on init .. end on for all 127
possible MIDI note numbers can could check which region does have a sample
mapped to a MIDI key:

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

And no need to make $lokey and $hikey polyphonic, as their values don't ever
change during individual note events.

> on note
> {message ("Note played is number: " & $EVENT_NOTE)}
> {uncomment for getting lokey/hikey instrument range numbers}
> 
> ignore_event()
> 
> 	if ( $EVENT_NOTE +2 > $hikey .and. $EVENT_NOTE <= $hikey )

".and." is a bitwise and, which is quite special. In this case you would just
use the regular logical-and without the dots ("and"). Not that it would make
much of a difference in this case though.

> 	
> 		$n := random(0, 2)
> 		
> 		select $n
> 		
> 		 case 0
> 		 
> 		  $RR := play_note($EVENT_NOTE -2, $EVENT_VELOCITY)
> 		  change_tune($RR, 200c)
> 		 
> 		 case 1
> 		 
> 		  $RR := play_note($EVENT_NOTE -1, $EVENT_VELOCITY)
> 		  change_tune($RR, 100c)
> 		 
> 		 case 2
> 		 
> 		  $RR := play_note($EVENT_NOTE, $EVENT_VELOCITY)
> 		 
> 		 end select

As $n is used nowhere else, you could also just inline the random() call like:

	select random(0, 2)
	  ...
	end select

If you want a true round robin behaviour, not a randomized one, then you would
probably use a global variable instead and increment it with each note:

on init
  { important: not polyphonic! }
  declare $rr
end on

function bumpRoundRobin
  synchronized
    $rr := $rr + 1
    if ($rr > 2)
      $rr := -2
    end if
  end synchronized
end function

on note
  call bumpRoundRobin()

  select $rr
    ...
  end select
end on

> 	
> 	else
> 	
> 	if ( $EVENT_NOTE -2 < $lokey .and. $EVENT_NOTE >= $lokey )
> 	
>   		$n := random(0, 2)
> 		
> 		select $n
> 		
> 		 case 0
> 		 
> 		  $RR := play_note($EVENT_NOTE +2, $EVENT_VELOCITY)
> 		  change_tune($RR, -200c)
> 		 
> 		 case 1
> 		 
> 		  $RR := play_note($EVENT_NOTE +1, $EVENT_VELOCITY)
> 		  change_tune($RR, -100c)
> 		 
> 		 case 2
> 		 
> 		  $RR := play_note($EVENT_NOTE, $EVENT_VELOCITY)
> 		 
> 		 end select
> 	
> 	else
> 	
> 	if ($EVENT_NOTE -1 > $lokey .and. $EVENT_NOTE +2 < $hikey)
> 	
>   		$n := random(0, 4)
> 		
> 		select $n
> 		
> 		 case 0
> 		 
> 		  $RR := play_note($EVENT_NOTE -2, $EVENT_VELOCITY)
> 		  change_tune($RR, 200c)
> 		 
> 		 case 1
> 		 
> 		  $RR := play_note($EVENT_NOTE -1, $EVENT_VELOCITY)
> 		  change_tune($RR, 100c)
> 		 
> 		 case 2
> 		 
> 		  $RR := play_note($EVENT_NOTE +2, $EVENT_VELOCITY)
> 		  change_tune($RR, -200c)
> 		 
> 		 case 3
> 		 
> 		  $RR := play_note($EVENT_NOTE +1, $EVENT_VELOCITY)
> 		  change_tune($RR, -100c)
> 		 
> 		 case 4
> 		 
> 		  $RR := play_note($EVENT_NOTE, $EVENT_VELOCITY)
> 		
> 		end select
> 		end if
> 	
> 	end if
> 
> end if
> 
> 
> end on

Alternatively you could also reduce the overall amount of code lines
tremendously by dropping the select ... end select approach and using
arithmetics instead:

	$n := random(-2, +2)
	$RR := play_note($EVENT_NOTE + $n, $EVENT_VELOCITY)
   if ($n)
     change_tune($RR, $n * -100c)
	end if

CU
Christian