Really good point! But I’m afraid there is no real great solution to that:
We could offer the old sample sets as a separate “legacy” library, but this may be problematic in cases of “questionable licensed” sets.
We could symlink new libraries previously available to an extra folder?
We could just state that this problem exists…
If this is technically possible we could have a similar user interaction message we might know from Kontakt (“files missing, please search for missing files or load an alternative sample set”)
I’ve one question again concerning this, taking the example of your (extraordinary) rhodes set. As far as I understand you have:
Looped (3c): Mono
Looped (3c): Stereo
Looped (3c): Both MIDI Channel dependend
Full Samples (3d): Mono
Full Samples (3d): Stereo
Full Samples (3d): Vibrato
Full Samples (3d): Mono with velocity crossfades
Full Samples (3d): Stereo with velocity crossfades
Full Samples (3d): Vibrato with velocity crossfades
Don’t you think we could reduce these? I understand looped versions are for storage size reduction, this makes no sense if we include both versions. The stereo version is algorithmically (by tuning the M-S sides). And the vibrato can be modulated in the sfz as well. So what do you think about using the 3d mono version and add the functionality in the script and/or leave that to possible effect engines? For me it’s just about the clarity of the menu (would be acceptable though if these files would go to a “variants” submenu).
This is one example for other future questions in the same direction.
I won’t delete anything from the table, and good point about SetBFree! I knew that but missed it completely. (This is why we discuss before we do.)
I think sources with questionable licenses are out.
That will rule out a lot of the current samplesets. I stand by my suggestion that we not include them but make it easy for the user to load them. That absolves Zyn from any legal issues, and all these samplesets have been floating around the Internet for decades.
Zyn needs a license target for distribution, and when that’s established we can adjust policies accordingly.
LMK what’s wrong with the table formatting. We may need more columns, of course.
I’d prefer to make it a lot easier, by pointing to the original sources (or our copies of them if necessary.) I’d really like it to be a “click here” operation, and I think that’s very doable.
Absolutely. We don’t need them all. We definitely don’t need the one where it depends on MIDI program.
For one thing, I can make the stereo width controllable by mod wheel. I’ve done that for 3d (don’t recall if it’s committed yet) and I plan to do it for 3c too. But mono samples are good when mono is preferred and one needs to keep the voice count down.
Note that I tried to emulate the stereo effect using SFZ but failed. If someone can find a good way to do that, great. Maybe I’ll try again.
We could omit stereo vibrato samples. That was fun to do and maybe someone likes it, but it’s an edge case. It’s not possible to do that kind of vibrato in the sfz file, unfortunately. (Even with the sampleset it’s not the normal type of stereo vibrato, but I like it a lot better than what we can do by sfz.) In any case, you can do a classic stereo vibrato with an audio plugin. Plus I can try again to do it via SFZ.
I don’t think we should care how many presets there are. For each sampleset, we should have a folder, and a default preset (.sfz file) per sampleset, allowing alternate .sfz files to be selected using the preset page. The default could simply be the lexically first entry.
But we certainly don’t need the image bloat due to the samples. I don’t think it’s necessary to include the unlooped sampleset by default. I play both sets routinely, and rarely notice or care which one I happen to be using. So, by default, we’d only have looped mono and stereo samples, which isn’t a ton of space.
It would need to include all the 3d-samples just directly into “Samples”, the versions are selectable by UI control
scripted vibrato is obviously nonsense when chosing the vibrato patch, but it’s just a proof of concept
there is an alternative, but commented version of just an amp tremolo
values are arbitrary and would need adjustment
By the way, the stereo effect you did to the stereo samples could probably also be achieved by sfz by making copies of the region, apply tune=2 to both and pan=-100 and pan=100 accordingly.
It can also be made more pleasing and wider when adding
lfo01_wave=1 // sine instead of triangle
lfo01_pan_oncc77=200 // wider
I haven’t had time to check out your sfz yet, but I plan to, and see if I can’t find a way.
I remember why I didn’t like left and right sides detuned slightly: it’s because this doesn’t cancel to mono without artifacts. And yes, I suspect the alternative of duplicating my mid-side approach would work but would increase voice count from two to three or four per note, which isn’t good for Zynthian, though not an issue these days for laptop/desktop. I suspect there was another reason, or I just didn’t manage to make it work even though it’s possible.
Regarding stereo vibrato, I can’t fathom why your approach wouldn’t be perfect. As I said, it’s not the traditional stereo vibrato (because the LFO triggers independently for each note) but my approach does that too, and I like it – learned the idea from my MR76, which had a lovely Rhodes that did that (and I wish I still had my samples of that but they were lost in a fire.) So, no idea why I felt I needed to bake it into the samples. Even if I did have a reason, I doubt it applies for Zynth.
Note that doing stereo vibrato on stereo samples won’t sound the same as on mono samples.
Anyway, with the looped sampleset we’re not talking about much data. But I’d definitely drop the stereo vibrato samples, regardless of whether we do something to produce a similar effect or not.
So, I do plan to make some changes in these areas, though it’s a lower priority than other issues at the moment. But I will check out your sfz soon, and thanks for it!
Can we agree that the UI should provide one instrument to choose per folder, and if the folder has multiple sfz files, those appear as presets? I think that makes things a lot clearer for the user.
Can we also agree about primary and secondary samplesets, where primary are pre-loaded and secondary are made easy for the user to load? The latter part is fairly simple except for the UI implications. For each secondary we’d issue a command. In many cases the command would be the same script that pulls all “similar enough” samplesets and provides the repo URL, and for any special cases we could have custom scripts.
If we find that a small set of conventions helps us manage Github repos, and if that requires changes to github/sfzinstruments samplesets, I believe I can usher those changes in.
One more proposal, and again the biggest trick is the UI, but it’d be nice to have an audio sample per sampleset. In many cases those already exist, and in other cases we can pretty easily create them. But, later for this! Anyway give it a bit of thought.
Generally, yes! We must consider that the upper folder structure also appear as “presets”, so we have to deal with another layer of the same preset structure, which would be categories (electric pianos, first level folder) → instruments (jRhodes, second level folder) → variants (i.e. jRhodes Mono, multiple sfz-files in second level folder). Another idea could be that the variant selection also appears as an addional auto generated rotary encoder or any other user input method inside a patch. Perhaps we could a short list view for stepped named parameters like these, so instead a rotary encoder presenting the currently selected value as a name it could present a tiny scrollable list inside the controllers UI real estate or as a short pop-up frame? I’d anyway still prefer to reduce its number as far as it appears meaningful. I would really appreciate a clear patch list were you find your desired instrument quickly.
I’m generally a big fan of curating as good as possible. I.e. if we have two grand piano sets with a distinguishable character, I’d just present both already, but omit everything that is of clearly lower quality than the primary one. We might not forget that for any further customization there is still the user soundfont bank.
For what purpose? We have the patch preload already, so you can review the sounds on the fly already in the selection list. Specifically for soundfonts I would revert back to preload on note-on though.
That would be the wrong way to show it. I’m assuming we have some UI changes. If we’re doing this without any UI changes, then we can take a different approach.
BTW, when you follow an insult with “kind regards,” you show what kind of person you are.
Hey, slowly! I don’t detect any insult here, please take into account that in a forum with a majority of non native speakers speaking english subtle distinctions of politeness in greating formulas might not be familiar to everyone.
The example @HansR gives is in fact the current form of presenting sfz for sample sets having more than one script per folder. I we add another layer, this would be fine.
I’m not aware of any insult. I think you refer to the Google translation of „In der Beschränkung zeigt sich der Meister“ and you don’t know Goethe’s sonnet.
