LMMS
SF2 GENERATOR!
Enhancement Summary
this feature will take VST to the next level! now instead of routing VST's through element just to load VST3, why not only have to do this once by trying our best to convert it into sf2
Implementation Details / Mockup
the most 'intuitive' way i build sf2 is by first exporting a .wav with a music scale (c1 to c8 or wather) in half steps with each notes being ~8 seconds ( as a multiple of 2 to be friendly with binary/hex), and then split the whole thing into separate wav each of ~8 seconds or whatever the interval is defined as.
there would be a residual folder with the instrument name and a lot of .wav's with the name of each note (ie. c1.wav, f#4.wav) alongside an .sf2 of the chosen instrument's name.
EDIT: alternate idea ============= "export as chromatic scale" (given a start and end)
it outputs a folder wich each note post fixed with .wav
==================
So you want LMMS to have a built in SF2 editor?
not strictly(as that would require a gui and all of that), but rather a way to sort of export a 'tone from vst' into a .sf2
Seems like a really specific feature, no? Polyphone is an already existing solution, and is also FOSS. You could create a project which plays notes you need (usually two per octave and a soft, mid and hard variant). Then you just save it as a template.
I think it's too specific, at least, but maybe some dev would want to implement it.
Seems like a really specific feature, no? Polyphone is an already existing solution, and is also FOSS. You could create a project which plays notes you need (usually two per octave and a soft, mid and hard variant). Then you just save it as a template.
I think it's too specific, at least, but maybe some dev would want to implement it.
that was the point of me updating with "alternate idea"
EDIT: alternate idea ============= "export as chromatic scale" (given a start and end)
it outputs a folder wich each note post fixed with .wav
==================
Upon further discussion, it seems as though OP specifically wants the ability to export every note from, for instance, the output of a VST, into WAV files that can be used without loading the VST itself. I myself think SFZ might be a better format to do this given the nature of the request. But since SFZ isn't supported, one is wont to ask for SF2 output instead.
Upon further discussion, it seems as though OP specifically wants the ability to export every note from, for instance, the output of a VST, into WAV files that can be used without loading the VST itself. I myself think SFZ might be a better format to do this given the nature of the request. But since SFZ isn't supported, one is wont to ask for SF2 output instead.
basically this
Mhm, I understood that, as well, thanks! I still feel as if this is just too much work for too little reap. You make one MIDI clip with those notes in a template or regular project, drag your instrument to it, export that and take it to a Soundfont editor.
We should aim for native VST3 support.
There's noone to make it, currently.
This feature is super cool, as it would help many people with low-resource PCs, but for that it would be better to use SFZ. For my part, I think they should first implement some SFZ player, and then this feature, since with the SFZ format it would give you the samples separated into folders and the .sfz file with which you can play them all together.
I fully agree with @bratpeki's assessment:
I still feel as if this is just too much work for too little reap. You make one MIDI clip with those notes in a template or regular project, drag your instrument to it, export that and take it to a Soundfont editor.
@bratpeki mentioned Polyphone, an OSS soundfont editor; here are links to Polyphone's website and source. I see no benefit to making soundfont generation a native LMMS feature when Polyphone exists for that purpose.
I think they should first implement some SFZ player
As far as playing SFZ files go, you can use the VST2 version of Plogue sforzando in LMMS. I do imagine adding SFZ support to LMMS's soundfont player would be valuable, but that's beyond the scope of this discussion.
It's 2025, we could be thinking about the reverse process: training AI to create patches in existing LMMS virtual synths, setting parameters to recreate the sound character of selected PCM samples sourced from soundfonts or music pieces.
This also, would be most appropriate to be implemented in a seperate project from LMMS.
I'd suggest this issue be closed; it sounds like general consensus that this feature is not something we are planning to implement.
