1) The mod. ratios, they do nothing.... ok, ok, so they actually do work, but I noticed that an irritating thing that is basically inherited from Motif (also being my MOXF). The minimum modulation amount in my opinion could be a lot subtler. Previous series, like CS and EX had it more shallower. This also happens with random pitch. So what about ratios? Well, set "part LFO" to say 32 depth, for active element set ratio of 8, and then of 9... tell me what you hear - because at or below 8 the modulation will not work - at all, and audibly "jumps" for higher ratios.

Thanks for your interest in the MODX.
Recommended Reading: The Extended LFO

Then you’ll be happy about this feature found in the MONTAGE/MODX. The old LFO range was 0 to about 42Hz, the extended LFO range is 0-1356.59Hz. This increases the values at both ends of the scale much more selection among low speed values and extends far into the high speed end as well.

2) The author of Polish translation of the manual should be fired. Out of a canon. Into the sun. 'Nuff said... Seriously though, the translation is flawed and convoluted. Luckily I always preferred English manual and google.

You should express your feelings to the regional distributor. That is where that information will have the greatest impact... and perhaps, get you some movement on change.

3) Am I right to assume that MIDI channels are "hardwired" to parts? So If I want to externally control voice on MIDI channel 5, I need to assign it to part 5? The manual doesn't specifically mention, and I haven't found any parameter to assign MIDI channels to parts, so I guess it's a correct assumption.

It is correct that Parts are fixed to the corresponding MIDI Channel, but to appreciate how this new engine deals with control you must take a deeper look. There is a fundamental difference from the previous models.

The Motion Control Synthesis Engine allows you to use and control multiple MIDI Channel simultaneously. On a previous model (Motif/MOXF) if you were playing four Parts (Choir, Piano, Strings, Pipe Organ) you would do so and would record that data on a single channel. When you then wanted to externally control the volume, you had the following options... all volume together, either all at the same value cc7, or in proportion to each other, cc11, or you could have some parts ignore volume control entirely.

The control matrix in the new engine allows more refined control on a per Part basis. The same musical control gesture can be used to change the four Parts’ output level and each will have its own unique programmable response curve. You can ‘design’ the application of the change of the four Parts so that each does exactly what you wish, at a rate of change and in the direction you define (including changing direction of change as you desire). Further, the change can be manual, automated, tempo driven, looped, or one shot. And can be done with a single control gesture.

In short, new ways to do control performance of sound are available. Next, add to this each Part of the MONTAGE/MODX Performance, can be a complete instrument (as in previous models) or you can use several Parts to construct a massive lay detailed instrument with enough segments to do exactly what you need. To get a handle on this, it is important to talk about the changes in the AWM2 engine found here. You are no longer limited to a maximum of eight Oscillators (Elements) per the instrument you might be constructing.

The XS/XF introduced XA CONTROL, a way the AWM2 engine could switch Elements to expand the range of articulations available to the performer. You could bring in a guitar harmonic, or pitch slide, etc. You could have repeated notes change slightly (or dramatically) in timbre to make it more natural sounding. You could have a mono legato phrase automatically/seamlessly switch to a sample set without a new attack portion. The MONTAGE/MODX allow the programmer to add additional articulation, additional velocity switching... a piano with 17 different Waveforms reconstructing the soft-thru-loud changes, an 18th Waveform to articulate Key-Off sound.

You can easily construct your sounds without limiting yourself to just eight oscillators. Say you’re in a band and you need to be the “horn section” (from the keyboard)... with a multiple Part MONTAGE/MODX Performance, I can assemble and *perform* standard full horn section, just the saxes, just the brass horns, sforzando articulation, a “doit”, a fall-off, a shake, bring in a bevy of high trumpets on-demand, etc. This engine allows you real time access to a variety of high-quality horn gestures, that can be assembled in one Performance.

You can control the equivalent of eight Motif XF (8 Element) Voices, simultaneously. Doesn’t mean they all sound together (they can) but the concept is to have a palette of sounds, a bigger vocabulary of musical articulation per the sound you are tasked with emulating. You determine *when*/*how* each articulation or segment plays. Assign Switches, Assign Knobs, the Super Knob, the MW, Foot pedals, velocity/note ranges, etc.,etc.

Now, in a situation where you are using multiple Parts to construct one massively flexible real-time playable sound, you want the instrument to generate MIDI messages so that every nuance of every performing gesture you make is accurately documented in an external recorder, like a DAW. This brings you to the point where you must consider what can be under your real time control in any Performance... you have eight Arpeggiators that can be used simultaneously - to interact and play musical phrases; you have 8 Motion Sequences available to automate parameter changes. And here we are not just talking about the ‘usual suspects’ (cc7, cc10, cc11, cc71, cc74, etc.) we are talking some 251 parameter destinations from the synth engine and onboard effect processors. You are not limited to just the cc messages MIDI makes available!

One huge paradigm change when dealing with “macro controllers” (the Super Knob and the 128 Assign Knobs), say you turn the Super Knob and simultaneously fade up three Parts, Pan four Parts from left to right, fade out another, and open the filter on the lead your playing... if you’re thinking I need to adjust cc10 on Part 5, so I need to send a stream of MIDI data on Channel 5 to alter how fast it is moving from left to right... STOP. That is not what you will be doing.

Remember, Part 5 is one of many Parts under control of the “macro controller” — if when the control gesture was created it moved
Part 5 too far, too fast what you edit is NOT the control gesture that made the change, you’ll want to edit the synth... you edit Part 5 and how it is responding to the control gesture. It has its own programmable DEPTH settings within its CONTROL SET. You can, too a dizzying degree, control *how* this Part responds to the existing control data. By editing the Curve, the Polarity, the Ratio, and Shape Parameter, I can adjust the distance the Pan moved and at what rate it changed position.

Because when multiple Parts are all on the same MIDI Channel, if you Pan one, you Pan them all — this is one option, and opting to ignore Pan is your only other option.... the new engine addresses each Part with its own direct parameter change message., these parameter change messages are assigned to a physical controller. You design each control gesture to precisely cause the changes you desire to each Part separately, the Super Knob can address across multiple Parts.

The Keyboard can Control as many as eight Parts simultaneously.
The Super Knob and it’s Common Assign Knobs can address all 16 Parts simultaneously.
How each Part interprets the incoming performance gesture is separately programmable.

Yes, you can use this system to affect changes on Parts being played from an external sequencer or controller, you can plug the audio of a slaved device coming in through the MONTAGE/MODX A/D IN, and you can treat that external device as one of your eight KBD CTRL PARTs... morphing to it with the Super Knob, applying Insertion Effects to it, controlling those effects in real-time, etc.

Thanks for the question.

Recommended Reading:
The Extended LFO
Then you’ll be happy about this feature found in the MONTAGE/MODX. The old range was 0 to about 42Hz, the extended LFO range is 0-1356.59Hz.

That is all well, but I'm not talking about frequency, but modulation depth, especially pitch modulation depth.

[...]XA[...], Motion Control[...]

OK, but it's not what I asked: I had tried programming exactly the same sound and hit the wall as with MOXF.

OK, so let me elaborate: A violin, or better yet, a cello - you usually play it on a single string or switch strings seamlessly, but often you can use two or more strings to produce a chord.
Now default MODX violins and cellos are polyphonic, when you play you get that ugly attack on every note. In fact, looking at the element structure, I can't even reach XA legato condition, and switching between poly and mono has no effect. I wanna change that. But there you have it: When creating new performance I can't use part mono/poly switch, because a) it's global for part;
b) in "poly" mode XA legato condition is never reached.
c) in mono mode, obviously I can't play chords.

Now try recreating the same effect with a rendition of Vangelis' signature brass sound... Believe me, I tried.