If I wanted the same I would start with the same exact envelopes. If I wanted shorter, I'd start reducing the appropriate time values. Or longer, I'd increase the time values.

That's the easiest way since factoring in level differences between two envelopes makes it difficult to do the math to resolve this.

The other thing you could do as a blind person is to take something that's easy to graph over time (like PEG or AEG) and see how pitch or amplitude changes over time using a oscope or similar tool. This would be the (kind of) tool to reverse engineer all of the gory details of envelopes although would not be necessary if you were able to make the levels match between the modulator and carrier. With AEG this should be doable. PEG seems more independent so copying say the AEG envelope would probably not yield whatever you're trying to get out of pitch envelopes.

More than trying to calculate the absolutes in the time domain, the documentation is just making you aware that you could have an AEG that's too short (in the carrier) to allow the modulator envelopes to complete. And that if you AEG on the carrier a short pulse then if you also wanted a frequency sweep (frequency sweep on the modulator) to happen over this short blip - to be sure to have the modulator use a quick FEG attack that more or less matches what the carrier's AEG is doing. Or even faster if you want to go up and down in the amount of time of the short audible pulse.

And also, to me "faster" would always start with the same attack level and time and then dial back the attack time to get faster.

I know there are limitations to this approach - but I hardly find instances where I need to get more exotic than using this kind of approach.

If I wanted the same I would start with the same exact envelopes.

My initial goals revolve around trying to understand what EXISTING sounds are doing - not in trying to create new sounds myself.

That was/is the thrust of my other thread
https://yamahasynth.com/forum/fm-x-q1-do-all-8-operators-receive-the-same-key-on-and-velocity-info-and-act-identically-
regardless-of-their-position-moderator-carrier-in-an-algorithm

the documentation is just making you aware that you could have an AEG that's too short (in the carrier) to allow the modulator envelopes to complete.
. . .
And also, to me "faster" would always start with the same attack level and time and then dial back the attack time to get faster.

The 'rub' is that what you say is EXACTLY what I am trying to understand.

In that other thread I showed this for a modulator in the MARIMBA performance.

Teaser: here are the envelope parameters for the op 3 modulator:

Attack Decay1 Decay2 Release
Time 95 35 42 98
Level 78 78 0 0

That attack time is so long I don't understand how it could possibly have an impact for a percussive-type
Mallet strike.

One of the things Bad Mister said seems to be in line with that

If the Modulator has a slower EG attack than the Carrier it is involved in influencing, then you can anticipate that the timbre will change over time if the KeyOn is maintained long enough for the Modulator to come into play.

When I create a carrier with the above parameters it takes several seconds before you hear any sound at all.

So taking BMs comment into account how could an attack time of 95 possibly be useful for a Mallet sound?

That is why my question in that other thread was whether the envelope would somehow act differently for a carrier
as opposed to a modulator. That was the only explanation I could think of for why a time of 95 was used.

At this point I'm thinking that modulator can't really be doing much of anything but was never removed/deactivated
because the times were so long it wasn't really getting the way of the final sound they came up with.

That's the only explanation for the long moderator times I have at this point. Unless setting the op 3 level to 0
somehow didn't really keep it from affecting its target.

If you just add up the values you get:

Op1 total envelope length - Hold + Attack + Decay1 + Decay2 + release
Op2 total envelope length - Hold + Attack + Decay1 + Decay2 + release

Adding the Time values alone is really not what you want to do… because the Time parameters are not going to necessarily give you what you need without factoring in the Level settings and Velocity…

Remember, in music there are several things that come into play… a staccato 8th note’s EG can behave differently than a held whole note. Playing staccato will execute the Attack Time and go directly to Release Time (skipping completely the two Decay stages. Music is made up of all different note lengths and also, different Velocities can cause different EG responses.

Remember, the Level determines how far the envelope travels before embarking on the next stage. The Note-on Velocity can also influence the movement of the EG segments.

Start with a simple Sawtooth Wave (one Element) and build a complex Amp EG
The farther the Level segment is from the last point, the longer it takes to travel to get there.

“Time” does not “add up” like its seconds… Time and Level should be adjusted by ear (not with a calculator).

The Amp EG controls loudness
The Filter EG controls the timbre
The Pitch EG control the Elements pitch

I'm not using MARIMBA off of a free DX7 converted library rather I'm using "FM Marimba" which is already a Preset and seems to generally match this Marimba "patch" you're working with.

Operator 3 has that long 70+ time on attack. And yes - this is much longer than is allowed by the carrier. The programmer (manny) didn't necessarily "care" about what's going on after the sustain gets cut off. The sustain dominates so this slow ramp was all I think was cared about.

If you take the decay and add some time to the carrier (6 I think it is) and also mute OP7/OP8 (I think ops 1-2 are already level 0 and effectively muted) then you'll be able to hear OP3's ramp complete. You have to crank up the decay time a lot to get the whole thing - but you can hear how OP3 is changing the frequency spectrum as it ramps up.

Then mute OP3 - you can prove to yourself that the frequency modulation you hear is due to OP3's AEG.

At any rate - in Marimba absolutely OP3 is getting truncated. I think this is all good.- it's by design.

When sculpting FM the world is a lot different than traditional synthesis and often you're exploiting how things work to achieve your end result. And often you would use algorithms to make little small changes to some section of the sound.

The goal isn't always to have the envelope complete in modulators before the carrier goes silent. Everything is a design decision with the freedom to make your own rules.

BTW: since BM sneaked in as I was typing - I'll add: I actually appreciate both approaches. Using your ears and running the numbers. I see value in both approaches. I do think the barriers to using your ear are lower than what it would take to reverse engineer a mathematical understanding. And although I appreciate and often prefer knowing the math - practically I don't worry about that in many instances.

If you dig, you'll find me asking the same questions regarding envelopes. Some information can be gleaned from DX7 emulation source code which is out there and others have looked at envelopes in much gory detail. I don't really have the equation/table/summary of how this works. And ultimately I gave up on this because eventually I need to get things done and paralysis through analysis can be an enemy to progress.

Adding the Time values alone is really not what you want to do

I didn't post it because that is what I want to do. My question was in reference to your statement that I quoted from the article about the relative length of two envelopes.

How, exactly, do you compare two envelopes to determine which one is 'shorter' than the other?

I am NOT trying to create my own sounds - I AM trying to understand sounds that others have created

Music is made up of all different note lengths and also, different Velocities can cause different EG responses.

I agree - but that doesn't shed any light on my question of how you compare two envelopes?

How can an attack time of 78 be at all useful in that Marimba sound I discussed that will surely come and go before the attack gets anywhere near 78?

That is what I am trying to understand.

Operator 3 has that long 70+ time on attack. And yes - this is much longer than is allowed by the carrier. The programmer (manny) didn't necessarily "care" about what's going on after the sustain gets cut off. The sustain dominates so this slow ramp was all I think was cared about.

Ok - I just used that sound and if I mute op 4 I don't hear ANY difference at all. So either I have really bad hearing or that attack time of 77 in op 3 is effectively taking op 4 out of the picture.

At any rate - in Marimba absolutely OP3 is getting truncated. I think this is all good.- it's by design.

Can you offer an opinion as to just what op3 and op 4 are contributing? Visual Analyzer shows little difference in the spectrums.

The reason why it doesn't matter is Manny is setting up a change in the frequency response and is using attack level + time to get at the desired result. Obviously, the peak of attack (the final level value) is not the goal. He wanted the impact to go from "here" (t=0) to "there" (t=carrier decays to 0) at a certain rate and is more concerned about the level when you hear the carrier chops off the sound and doesn't care about what happens after that.

So for this kind of thing, one would be interested in the starting level of AEG and then some "slope" which matches how fast they want the frequency modulation to happen. And absolutely this can be tuned by ear. There's probably a lower level with a different time value that would not get cut off but still creates the same "slope" or something close to it. Then the attack, at least, wouldn't be truncated but the end result would sound the same.

There's probably a lower level . . .

So your answer to this question is NO?

Can you offer an opinion as to just what op3 and op 4 are contributing? Visual Analyzer shows little difference in the spectrums.

I don't need to offer my opinion about OP3 - I already gave the way. First Mute OP7,OP8. Crank up OP6's decay1 time so the envelope stays non-zero for longer. Then listen to what you hear that's slow. You'll hear a slow change in the frequency response (kind of like a cutoff sweep) and it's slow because OP3 has that slow ramp. Then mute OP3 to prove to yourself that OP3 was the thing doing this. This seems like deja-vu.

And what you will clearly hear when OP3 is not muted and the carrier doesn't chop it off - you'll hear what the frequency response is doing over a longer period of time if OP3 was allowed to "complete" (at least complete the attack phase). And then you can understand from this wider picture that a shorter section of OP3 will change the frequency in the same way but just over a smaller section of the sound because the sound ends before OP3 gets through the attack phase.

I've offered in pretty good detail what exactly OP3 is doing. I can move on to OP4 but the basic approach is the same. Mute the right stuff - listen to what having OP4 does by muting it and unmuting it.

And that lower level thing is just if it matters that say the attack completes then you can probably come up with a lower level (and time adjustment) that will be the same slope as we have here but since it doesn't go "all the way" to the current level will end the attack phase (OP3) before the carrier goes silent. I don't see any value in this - but if it bothers you, there's a way. And then none of the other phases really matter because they aren't in the picture of the final result.

This seems like deja-vu.

I've already described how I did a test by copying op 3 to a carrier.

Crank up OP6's decay1 time so the envelope stays non-zero for longer.

My test eliminated op 6 and its decay 1 time altogether.

That allowed me to hear, and see with Visual Analyzer the path that was being taken. And it was so short it didn't have time to do anything that would have really altered what the original op 4 target was doing.

Mute the right stuff - listen to what having OP4 does by muting it and unmuting it.

I already described that I did do just that. Both listened with my ears and with Virtual Analyzer. No significant difference.

I'll just assume for now that you don't know what the op 3/4 combo is addig, if anything, and let it go at that.

Looks to me like it might have been added during the initial sound design and then abandoned when it wasn't needed.

Too bad I can't ask Manny himself. Because I'd still like to know either that it isn't doing anything or would like to know how/why the op 3/4 was added to begin with.

You're not doing the right thing if you cannot hear what OP3 is doing by following my suggestion.

I'm patient, though. Here's a more detailed approach:

1) Recall "FM Marimba" (because anyone with a MODX/Montage can do this too without jumping through hoops)
2) Edit and OP7 and OP8 levels. Change them to 0 to null these out. Here in practice I use MUTE but Mute gets reset under different conditions so zeroing out OP7 and OP8 is "easier".
3) Edit OP6's level screen and change the AEG presented there: Change Decay 1's time from 56 to 99
4) Press and hold middle C

You should hear in about the 1st 3 seconds the sound of something ramping up then going away. A "sizzle". This sizzle you hear is OP3

5) Mute OP3 (you can use mute this time or also set the level to 0. Use the [EDIT] (COMPARE) button to see the old value if you set it to 0. You won't hear any of that sizzle stuff anymore. That's because OP3 was causing that.

And now you appreciate what OP3 would do if it made it all the way up to the attack level you can infer what it would do if you chopped that off early.

OP3 is adding a kind of (non-percussive) "zing". Downstream other operators contribute to the percussive-ness of OP6's (carrier's) output.

OP5 adds the mallet strike. OP4 and OP5 are almost identical in the level envelope but contribute differently to the sound even if you turn OP3 off (level=0 or muting). The reason for this is OP5 is a 4.0 ratio and OP4 is a 0.5 ratio in the Form/Freq menu. So OP5 is going to excite higher frequencies while OP4 is much lower. OP4 by itself hardly sounds like anything but if you listen closely you'll hear that with OP4 off the sound has less mid frequencies (sounds duller with OP4 off + OP3 off + OP5 off so you don't hear that mallet strike stuff either - vs. OP3 OFF/OP4 ON/OP5 OFF which is brighter in the mids).

Really the main purpose of OP4, to me, is to deliver OP3 to the party. The sizzle part.