Not sure if this is the sort of thing you're after.

https://clyp.it/sns3fe0x

All of this audio has no keyboard intervention from start to finish and is just a few "random" things affecting each other to change pitch of a sine wave.

Nice!

Yes, that is the concept of what I am looking to do. I want to play a chord and have it evolve over time by having different things modulated.

I am new and still learning this beast but is it possible to modulate modulators?

I am new to FM so still experimenting and reading.

How deep is the Montage modulation matrix?

Thanks.

From what I can tell (not having experience with the genre / concept in my own musical expression) - generative music is similar to 12-tone in concept. 12-tone music says that you have 12 different unique pitches (A, Bb, B, C, Db, D, Eb, E, F, Gb, G, Ab) - so come up with a composition that plays all 12 tones without repeating a tone within the sequence of 12-tones. Regenerative music says you have 127 "steps" of MIDI - so for whatever parameter you are trying to target (like note pitch, resonance, cutoff, etc) - to cycle through all 127 values without repeating any values.

Obviously a ramp from 0 to 127 would satisfy this - but the general idea is more along the lines that there is a more random order. Although 12-tone music would be satisfied with a chromatic scale as well - it's not the most "interesting" of results.

With limitations of sequencers - some have used a matrix where they pre-arrange a sequence of 127 unique values. With Montage - you cannot exactly do this because the user curves only give you 8 steps to dial in a specific value. So even here, at least using the Montage interface, you cannot get "pure" generative.

What I've done is just pick one of the many ways to get a semi-random result.

My goal was only to modulate pitch as the target. You can pick a different target parameter. In order to do this I:

0) Started with the Init Normal (FM-X) performance which is just a pure sine wave and never changed this fundamental shape except for pitch. I did make one edit to the PART (FM-X) - I changed hold *LEVEL* to 99 so the pitch would infinitely sustain.

1) Setup an MS Lane with the pulse type "Sequence" - which is fairly random sounding. This was MS Lane 1. The tempo I set to 5 (slow) and set the sync to tempo with a high unit multiplier (6400%)

2) Setup a motion control with source as MS Lane 1 and destination as pitch (FM OP number 1 frequency). Further "randomizing" was done here because I used the "M" curve type with bipolar polarity. Even a ramp would have caused some random sound since sequence was already "random" itself.

... This setup some random sounding pitch fluctuation.

I wanted to, from the get-go, use envelope follower as a way to add to the randomness.

3) Since envelope follower deals with sensing amplitude - I wanted to fluctuate the amplitude then feed this into the envelope follower then take the result and feed it back into pitch. So I setup an LFO with S&H (sample and hold - old school random) and had the LFO controlling amplitude with a reasonable amount of depth in order for the envelope follower to move.

4) Setup PART 1 envelope follower with the least setup/hold hysteresis so the "needle" would move around faster and setup the gain so it was around some area of "interest". Really all of this fiddling is random itself and doesn't matter what the details are exactly. If I could have the needle jump from 0-127 willy-nilly I'd do that. But this is harder to do and I would have setup a 2nd PART (not sounding) to do this rather than use the sounding part for this. Even so, it would be complicated to realize this.

5) I used motion control again and had the source as Envelope Follower 1 and destination again as pitch (OP 1 Frequency). After adding this, now there are two sources simultaneously affecting pitch as the destination.

I chose FM-X because it's the only way I can setup a sound to infinitely sustain after releasing. AWM2 doesn't allow for "infinite" hold times (by setting the hold level to a non-sloping one).

More interesting sounds can be accomplished by changing from the default sine wave to something else. Adding some form of an ARP can add some depth although there is no ARP here and no keys being pressed - so not altogether necessary.

You modulate modulators by choosing a SOURCE which itself can be modulated. That's why I picked envelope follower as it can be modulated by amplitude output of the PART it is following.

Note: I didn't answer the question on Modulation Matrix Depth because I'm fuzzy on your definition of what this is. Defining these would help - or even giving a modulation matrix depth "value" of some other keyboard (or plugin) you know this information for would help arrive at a definition.

Note2: if you want to target a lot of parameters in each PART - then using superknob automation with assignable knobs as the destinations would be a way to go. Then you would have 16 possible parameters per PART to modulate using a single "trunk" source (superknob automation) with varying ranges/directions of each branch source (assignable knobs). 16 per part is a resource limitation on the destinations.

This 16 destination limit can become cumbersome in the FM-X world if you have many operators (say all 8) and want to individually modulate each operator with a different "pattern" (curve). You have to use a different destination for each curve - so if you want pitch and level as two parameters that are modulated using different curves for each operator - then you've just used up all 16 destinations. Yes, the pitch (for example) destination does allow for modulating all 8 operators at once - but you only get to change the "depth" for each operator. So they would all be moving in the same direction - just with a different "multiplier" with only related movement. As Yamaha watches how keyboard programmers use motion control - it would be nice to eventually expand the resources in this department. For now, you can make some choices on how to use the 16 you have most effectively.

As a side-note: limited resources is not necessarily a negative. Take an analogy of ASCII art. http://www.ascii-art.de/ascii/ This would range everything from the smiley face ASCII two-character "pictures" we would do before the word "emoji" was coined - to photo-realistic pictures using different ascii characters. The resource is only around 100 characters (around 200 if using 8-bit) yet some results are so-so and some can produce true artwork in this medium. Same with synth limitations - using the limited resources some can churn out massive ear candy while others can only produce a handful of same sounding bleeps using the same tools. Working within the limitations is part of the "fun" of programming a synth.

Wishing/advocating for more resources ("the moon") is part of the "fun" of being a consumer.

Thank you for the very detailed explanation of your patch. Gives me a lot to work with and experiment with. Can't wait to get on my Montage later and experiment. I have yet to dig deep on this monster as I have been mostly editing the presets into different sounds. Time to get busy with some generative and randomness!

Cheers.

Since I mentioned "FM-X" I sensed some hesitation. But notice that I only changed a single parameter (Hold Level) for Operator 1 (which is the only operator programmed to make any noise in the initialized FM-X performance). There's not much to learn about FM-X in order to accomplish this since this is the familiar ADSR side of FM.

I intentionally kept the basic sound simple since you were more interested in the manipulation of sound rather than the core sound itself. The core sound you can play with at some point or use one of many presets available inside Montage or on the net (DX7 "patches" in sysex form and converted using DX7->Montage online converter).

If you have more than one carrier - then you may want to modulate all carriers (in the example I was using of carrier pitch). This is where you have the choice to stick to more simplistic single-carrier FM programs. If you do not know what a carrier is - check out the tutorials on FM synthesis in this site.

Point of order:
Calling 16 Source/Destinations a “limitation” is kind of absurd, just FYI. If YOU limit yourself to one Part I guess you can make such a claim, but building sounds in MONTAGE are not limited to 8 Operators... “the ‘moon’” indeed.

Infinite Sustain in the AWM2 engine is a matter of understanding the technology... sample-based. So you don’t Sample a sound indefinitely, that trip to the moon would last forever. Samples gain endless immortality via the Sustain function... which can latch the audio in its loop cycle. Designed to emulate music tones and musical instrument sounds by using audio recordings of real world sounds. Setting an AWM2 wave to ignore Note-Off will allow you to leave a sample infinitely sounding (looping) for use as a Modulation source or as an audio source. (And that’s a fact). Now perhaps you only meant normal programs, with normal EGs... but why “limit” it to just those conditions?

Also remember, literally any audio source can be used as a Modulation source via the A/D Input or via Side Chain Modulation (including the Vocoder), again it can be a silent modifier or you can route its Output to the main Outputs, if you desire.

Great discussion and thread, otherwise. I don’t want to be a “Debbie-downer”, it’s just that it’s hard to standby when it’s stated that there is some “limitation” rather than an admission that perhaps no one has found a way yet to do something. Proclaiming a limitation limits others from perhaps exploring on their own. Moon requests are fine, we can deal with crazy requests...

A lot of what Yamaha has been doing in the last 30 years is extending and expanding what can be done with sample based data in the performance of music (overcoming many of its built-in limitations)... FM-X, based on a digital technology that has its origin at about the same time, takes a completely different approach to synthesis.

@Chris: Can’t wait to download and hear the examples you’ve posted.

@Chris thanks for the examples, I really like the ambient one. Oh Cool I can try Montage connect, thanks for that.

@Bad Mister thanks for clarifying in detail. Having owned many synths like I am sure most of us have, I must say that I am highly impressed with the Montage and the reason I know this is because I can't stop playing it and experimenting with it. Once I power it on I wonder off into the world of sound and I am there for hours.

Cheers everyone.

The AWM2 comment wasn't a dig. I was explaining why I employed FM-X for this kind of thing since it's so easy to create "drone" type sounds with a simple one-parameter edit. Using the middle of the sample with loop points you can hold a note "forever" - but it's more to setup so suggestion went to FM-X. There was absolutely no request to change anything with AWM2 or anything negative. I felt it necessary to explain the "why" I went to FM-X to express one of the many areas where FM-X is the "better" tool for the job (gets the OP up and running quicker). The only request made here was regarding the future of destinations (count or pooling).

Regarding limitations ...

I didn't say the sky was falling - even mentioned that limitations are part of the experience/challenge/joy of owning any programmable device. Someone who wants simultaneous parameters to be "organically" (meaning many patterns which do not seem to sync with each other) modulated by a motion sequence is more likely to run into the 16-per-part limitation than some other uses - so I highlight this as a wall so there's no surprise. Hopefully, certainly, that wall is never met by the OP - but it's useful to know in this context.