Using play_note in a way that gets picked up by record

Hey there! While there are a few ways to achieve what you're looking for, I'd suggest going with the simplest one.

When bouncing recordings, the noise sample won't be audible if the noise slider isn't set to 1 (I'd change it to noise_slider # 0). To avoid delving into other solutions involving some advanced KSP, you could map it to a note that's outside the instrument's range, like C-1. Then, whenever the C-1 note is played, the noise sample will be triggered. You can also use a custom event_id to trigger that specific sample infinitely. Check this out:

on init 
   declare noise_id
   declare const NOISE_NOTE:= 12 {C-1}
end on

on note
   if EVENT_NOTE = NOISE_NOTE

      noise_id:= play_note(NOISE_NOTE, EVENT_VELOCITY, 0, 0)
      set_event_par_arr(noise_id, EVENT_PAR_ALLOW_GROUP, 0, -1)
      set_event_par_arr(noise_id, EVENT_PAR_ALLOW_GROUP, 1, 69)
    end if
 end on

Having the sustain involved could complicate things further. In my scenario, the noise_slider callback should be removed, and only the code I sent you should be placed in the relevant callbacks. However, since the sustain pedal will silence all active voices, you'll need additional code to protect that noise sample. This means you should write your own sustain pedal script within your main code.

Regarding the bounced audio being captured within a DAW, wouldn't it be simpler to just add the noise sample in another audio track?

If you prefer not to delve deeply into KSP for the noise sample, and considering it's not a major functionality of the instrument based on your initial post, wouldn't it be better to add that noise sample to a separate Kontakt instrument? You could then load both instruments (main and noise) within a single Kontakt instance. The secondary .nki could include a line that disregards the default sustain pedal behavior and simply ignores any sustain engagement, meaning that your noise sample imported to that secondary .nki will play infinitely.

on init
   set_condition(NO_SYS_SCRIPT_PEDAL)
end on

Firstly, you'll need to declare a custom event for your noise sample, such as noise_id. Then, create a function to trigger that noise sample or group whenever the noise_slider is greater than 1. However, you may notice that this could trigger more than one voice whenever the noise slider is nudged. However, you mentioned that you're limiting the played voices using the group settings.

Alternatively, you may need to script a sort of protection measure to fade_out the event and trigger it again. To trigger it again, you might need a flag indicating whether the noise sample should be played or not. Hope that helps you out! :)