You cannot extract the waveforms from the S90. If you do not find matching waveforms then you will be left with either living with "close enough for jazz" or go through the work to sample your S90 which is an art unto itself. If you want to get the original waveform as clean as possible then effects need to be all turned off. Your envelopes need to be set so the envelope (amplitude, filter, pitch) so that these do not alter the waveform. EQ should be flat as not to color the sound. Anything else that would alter the sound would need to be adjusted to a "benign" setting. You'll need to discover the best you can where the key samples are. Where the "root" sample is for a range of keys. And you only need sample those.

Then you'll need to make sure, on the other end, you assemble your samples properly. There may need to be adjustments made. The keymap needs to be constructed properly.

It's not such an easy task. There is software to help automate these things - although the best setup for even the automatic software would involve doing many of the same manual steps. If you choose full automatic mode then you may end up using more memory than you would otherwise need for the samples -- or you may tell the software to sample less keys which would result in some quality change.

... and -- in addition -- you ideally need to know something about the velocity layers of the original sample(s) so you can most efficiently capture and reproduce those. Again, software can help automate this but full automation may be a rough approximation that takes up more memory than necessary.

The software I am alluding to is "SampleRobot" - although it's not the only game in town for assisted sampling. The SampleRobot software doesn't assemble the waveforms for you - so there are still steps to take after this program produces its output.

Thanks for the quick and informative reply, Jason! I took a look at the SampleRobot you mentioned, and I don't think for me it's worth the $100 and steep learning curve to try to do that. Although, I might get lucky since the ModSaw waveform I'm most interested in might not have as many layers as (for example) an acoustic instrument waveform. Or, at least for my purposes, I might be able to take some shortcuts and still be satisfied.

If I were to try to capture samples manually, I assume I would just take the PHONES output -- using a custom voice to get the waveform as "uncolored" as possible -- and record it into stereo WAVs? What would I then use to turn those samples into a waveform usable by the MOXF? I've seen mention of a "Waveform Editor" software? Could you, or someone else as knowledgeable, point me in the right direction for that?

Finally, am I able to do this with the MOXF as my target synth, or is this one of those reasons people buy a Motif (or Montage) instead?

EDIT: Update. After auditioning the waveforms side-by-side using initialized voices, I've found the MOXF "1345 Med Detuned 3osc" waveform is super-duper close to the S90's "379 Wv ModSaw"! And as for the other waveforms I'll leave behind on the S90, I'll focus on looking forward instead of looking back. Thanks again for this forum; oftentimes a conclusive "no" is just as valuable as a "yes".

Glad you found something "close enough for jazz". That's what I did when I was translating Motif ES Performances into Montage. It's similar as the process of going from the S90 (Motif Classic era) to the MOXF. Same general time-span. I found a lot of "exact" matches and some weren't. So I fudged and got a new/different sounding result. Good enough. The audience doesn't mind and now I'm accustomed to the "new" sound more than the original.

If you really wanted to know Yamaha's opinion of how it translates - the ES series imports Motif Classic and translates according to Yamaha's best-fit matching. You'd need the presets in .w2e format.

Skipping a few steps ... They are in w7e format: http://www.motifator.com/index.php/support/view/motif_classic_voices_in_motif_es_voice_editor_format/

I assume somewhere in there is a voice that originally had "ModSaw". You'd look at that and see what the editor tells you is the waveform.

Docs do not show which Voices use what Waveforms (that'd be a bit much) - and there aren't any editors for the Motif Classic series still compatible with my OS - so I can't see if those would tell me anything. It doesn't matter much since you've moved past this. And how the ES or XS would have translated Motif Classic may be besides the point if it doesn't have such a great match.

Get back to focusing.

Even though you've hopefully let all of this go - I still couldn't leave well enough alone. Yamaha provides "lookup tables" when translating a Motif Classic era waveform (such as the original S90) to more more modern generation of instruments.

What I did was took W7E file I linked above which already had translated the waveforms in one step and imported it into the Rack-XS editor so I could see what Rack-XS would say is the new "translated" waveform name for what began as ModSaw in the Motif Classic era.

Here's where we are coming from - I had to figure out which "Voice" had the ModSaw waveform because "Voices" are what we have to deal with as the first layer of the onion.

You see that "SimpleSaws" has element 1 as "ModSaw". Great. BTW: I had to jump through a lot of hoops to get the MOTIF(classic)-RACK voice editor to work since it doesn't run on any modern operating systems.

... and then I import "SimpleSaws" (from the W7E translated classic preset file) into the XS's editor and inspect that voice:

... and above is the conclusion. Yamaha seems to feel that, at least in the Motif XS, that "OB Mod Saw" is the closest fit for what was once "ModSaw".

That doesn't necessarily mean your MOXF waveform you selected is "wrong" - just that someone who went through the automated steps to convert from Motif Classic era to XS would end up with "OB Mod Saw" - a waveform that does exist in the MOXF. Therefore, such steps in translation eventually to MOXF would have also yielded "OB Mod Saw" since - although I could be wrong - I'm not sure Motif XF (and I say Motif XF and not MOXF because of the "big brother" relationship - where Motif XF may have more files it translates) directly reads Motif Classic files and does a more direct translation.

Your ears ultimately dictate what's best. I just couldn't leave well enough alone and "had" to provide the translation-table perspective.

With a GB or more of user sample space in the MODX/Montage, and the old sample sets being so small in comparison, it would be nice if Yamaha provided all those old "missing" samples from the Pre-Motif-XS sample-based workstations (Motif ES, Motif, S30/S80) as something that could be downloaded into that space, to better facilitate people being able to bring in their old "legacy" sounds... ideally they could provide the Voices all converted as well. I posted that a while back on ideascale, if anyone might like to upvote it... https://yamahasynth.ideascale.com/a/dtd/S30-S80-Motif-Motif-ES-expansion-packs-for-Montage-MODX/274368-45978

I don't disagree. The XF at least still had some of the Xspand, Inspiration, etc promo sets that were compatible with it to bring-forward Motif Classic (or later) content that was left behind. There were a few promo packs for Montage/MODX - but compatibility issues plagued what, at first, seemed like the possibility. I haven't gone through the promo packs to know that "ModSaw" was or wasn't a part of all that. I'd think the searching would have uncovered it - but I didn't specifically look. Xspand Your World, for instance, doesn't list out all of the waveforms - just voices. And there's a lot of stuff in there with overlap to the "Inspiration" set.

One comment here - is that this thread is about MOXF - the Motif XF era "MO" feature-reduced version. So the MOXF of the OP has a few more options to bring forward legacy sounds with MOXF's compatible libraries (that is, if a flash card is in the box).