MODX8 Polyphony Issue - HELP -

[ITB-Music]

Hey Joshuaedo, welcome to the forums!

It has been covered in a number of topics already, but basically, it is NOT a bug. One way or another you are hitting the 128 voice polyphony limit. So, to figure it out you should share more information about how did you setup this performance. Do you play all 4 parts at once? If yes... you should be aware that strings and pads played with sustain can eat your polyphony fast. You should consider turning off the sustain for these parts and eventually raising the Release (as in ADSR) for these parts so they don't cut off unnaturally.
Anyway, share some more info, you can also upload your performance to Soundmondo and share it here and maybe someone will figure out how to optimize it.

[Marquest]

Hi there! I bought MODX7 about a year ago and still wondering what the hell is wrong with it's polyphony architecture. Same thing – stacking some sounds I need to use in my live performance with a band (e.g. Piano +EP+couple of pads) leads to heavy polyphony issues (piano notes dropout etc). I tried to use "single part" sounds only, choosing sample or FM engines wisely, but the result is still not satisfying. That's weird because I have KORG M3 and it's smaller brother M50 in my use and I've played them a lot on stage, stuffing it's "combi" layered patches with TONS of sounds (8 and more complex sounds for one patch sounding at the same moment), but I can't remember any polyphony problems at all. So funny because that old M3 has sick 120 polyphony limit versus 128-note AWM2 and 64-notes of FM-X on MODX. Is it something about MODX architecture???

P.S. The only solution I found is to use sounds that use less elements. But it looks a bit complicated because it isn't indicated anywhere and I have to assign the sound to my performance, go to "edit" and examine the elements panel.

[tomsargentgcs]

I experienced this very thing. I have an MODX7. I believe when you add a layer or voice, it uses polyphonic whether the volume is up or down. I added strings and started losing sounds. I deleted the layer and the problem when away. So I play at church and wanted piano, pad, organ and strings. I had to pick voices that were only using a single channel. I love the keyboard but this was very disappointing. I also had to be aware of how long I was sustaining the sounds like when playing pad sounds and allowing them to sustain for a long time. I have even tried to reset back to factory settings and it didn't seem to help. I would love to find a local MODX guru to walk thru this. I love that I can layer sounds and that this keyboard is so reasonable in price but when you hear you're at the polyphonic limit its very frustrating. It's also frustrating that I can't use the better richer piano voicings because they use so much more polyphonic. Does the Montage have this issue? BTW the specs for the MODX states 192 polyphonic and not 128.

[Saul]

The MODX has a total polyphony of 192 but that is split between 128 AWM2 and 64 FM.

I am not sure what is going on with the polyphony issue but it doesn't seem right in the year 2020 that we are still seeing this issue

I remember back in 2005 when the Alesis Fusion came out and you could layer anything you wanted without a hint of polyphony issues. 15 years later we are discussing note dropping on Yamaha's newest mid-range keyboard

Perhaps if they had used modelled instruments rather than samples we would not be where we are?

I don't know what the solution is other than to limit yourself to single element voices which kind of defeats the point of having a keyboard like the MODX in your rig?

[javelin276]

Yes, Yamaha got fancy with all the new voices in the MODX and went multi-part on a lot of them. They sound great, but they eat up a lot of polyphony. I have better luck doing multi-instrument multi-channel pieces when I use my Motif-XF, all of it's voices are single part. Yamaha should increase the maximum polyphony by a factor of 4 or more in the MODX so it can keep pace with the Motif it replaced.

[anotherscott]

The only solution I found is to use sounds that use less elements. But it looks a bit complicated because it isn't indicated anywhere and I have to assign the sound to my performance, go to "edit" and examine the elements panel.

Yeah, it is unfortunate that there is no indication (on screen or even in the PDF data list) of how many elements a sound uses. Though even then, it's tricky... a sound can have 8 elements and still only use up two instance of polyphony, if it never triggers more than two elements at a time (e.g. if it triggers different elements at different velocities or over different key ranges),

I experienced this very thing. I have an MODX7. I believe when you add a layer or voice, it uses polyphonic whether the volume is up or down.

Yes. That's why, for example, you could have the strings volume down, play a chord, and then fade the strings in. The only way that can work is it it were actually "playing" the strings (too low for you to hear) all along, so that it was "there" to be faded in. If you're not going to need to fade a sound in under previously held keys, then as I mentioned in the other thread, don't merely bring the volume down, but actually mute the sound. Then you can use Scenes to switch among your sounds, and the ones you're not playing won't use any polyphony.

Yes, Yamaha got fancy with all the new voices in the MODX and went multi-part on a lot of them. They sound great, but they eat up a lot of polyphony. I have better luck doing multi-instrument multi-channel pieces when I use my Motif-XF, all of it's voices are single part. Yamaha should increase the maximum polyphony by a factor of 4 or more in the MODX so it can keep pace with the Motif it replaced.

Every "single part" Motif XF Voice also exists in the MODX, so you can still layer those sounds as much as you could before. The multi-part sounds may sound better, but sometimes in subtle ways that shine when played solo, and may not necessarily be so noticeable when layered with 3 other sounds. So the strategy could be to stick mostly with the single-part sounds when doing lots of layering, and save the multi-part sounds for when a single sound will be more "exposed."

Even when layering single-part sounds, the MODX still offers numerous significant advantages compared to doing the same thing on its MOXF predecessor (a more fair comparison than to the much more expensive Motif XF which is more comparable to a Montage)... you get benefits like seamless sound switching, touchscreen patch selection, the FM synth engine, control of up to 8 MIDI parts instead of 4, effects on up to 12 parts instead of 8, superknob/motion control. And particularly when you factor in that it comes with the equivalent of the 1 GB flash memory expansion built in, it's a better value than what it replaced, as well.

[Stevie18]

I think anotherscott nailed it. Volume down does not help because in theory you could fade in and then you would expect the voices to sound as if they were triggered at key down. So being aware of multiple parts is the key to avoiding cutoffs.

Still, I have a MODX and a Montage, and they behave differently for the same (copied) sounds. The MODX starts cutting off earlier than the Montage. That’s for pure AWM2 sounds, where in theory the polyphony should be the same. I cannot explain that except that polyphony on the Montage is actually more than promised - or MODX lower...

Modeling would not improve I think because it usually is more processing intensive.

[Saul]

Modelling is not a processor issue these days. Well not unless you are using ancient cpu's and memory! Surely Yamaha would not be doing that right

[Macke]

Still, I have a MODX and a Montage, and they behave differently for the same (copied) sounds. The MODX starts cutting off earlier than the Montage. That’s for pure AWM2 sounds, where in theory the polyphony should be the same. I cannot explain that except that polyphony on the Montage is actually more than promised - or MODX lower...

This is indeed a interesting observation. The specs sheet would suggest that both boards should be able to handle min 128 AWM2 notes polyphony but the MODX is for sure scaled down in the sence hardware (one SWP70 tone generator vs two on the MONTAGE according to https://sandsoftwaresound.net/yamaha-modx-inside-stuff/ ) so the MONTAGE is more powerful as expected.

If there only was a way of seeing how many notes are used. Anybody with the service manual at hand that can check?
EDIT: One way could be to create a performance with a number of sustained parts active to easily reach the limit by pressing enough keys and just calculate.

This is pure speculation on my part but there might also be some unforseen sw issue/bug introduced that make the polypony less than 128 in some circumstances. Without a method to check what is used then it might be quite hard to actually prove this and request a fix.

/M

[anotherscott]

I have a MODX and a Montage, and they behave differently for the same (copied) sounds. The MODX starts cutting off earlier than the Montage. That’s for pure AWM2 sounds, where in theory the polyphony should be the same. I cannot explain that except that polyphony on the Montage is actually more than promised - or MODX lower...

This is indeed a interesting observation. The specs sheet would suggest that both boards should be able to handle min 128 AWM2 notes polyphony but the MODX is for sure scaled down in the sence hardware (one SWP70 tone generator vs two on the MONTAGE according to https://sandsoftwaresound.net/yamaha-modx-inside-stuff/ ) so the MONTAGE is more powerful as expected.

If there only was a way of seeing how many notes are used. Anybody with the service manual at hand that can check?
EDIT: One way could be to create a performance with a number of sustained parts active to easily reach the limit by pressing enough keys and just calculate.

This is pure speculation on my part but there might also be some unforseen sw issue/bug introduced that make the polypony less than 128 in some circumstances. Without a method to check what is used then it might be quite hard to actually prove this and request a fix.

Here's another theory of pure speculation. Whether due to different processing power or other architectural differences, maybe the note stealing algorithms between the two are different, such that a dropped note due to insufficient polyphony is more noticeable on one than the other. The simplest algorithm for what to do when no polyphony is available to play a new note would probably be to drop the oldest note played (first in, first out)... which is not necessarily the note of least noticeable impact. A more sophisticated algorithm could (for example) take into account the volumes of the already-played notes, based on both their initial volume level and how much the volume of those notes may have already decayed (due to their envelope settings) by the time a note theft needs to occur. There are other factors that affect potential noticeability of a dropped note as well, e.g. how the keyboard tracks multiple reptitions of the same note (with the pedal down), or even how a note is harmonically related to other notes playing the same sound. I wouldn't expect differences between Montage and MODX in these areas, but it's not theoretically impossible.

Modeling would not improve I think because it usually is more processing intensive.

Modelling is not a processor issue these days. Well not unless you are using ancient cpu's and memory! Surely Yamaha would not be doing that right

Modeling is a catch-all phrase for deriving your result through mathematical calculation. It can require small or large amounts of processing power, depending on what you're modeling, what your goal is, what the defined algorithms are. There have been modeled Hammond organ recreations for a long time even on relatively minimal hardware... it's relatively simply since the basic sound essentally consists of a combination of nine constant sine waves at variable volume levels. By comparison, there are, I believe, only three serious attempts at modeling piano currently available (Roland, Viscount, Pianoteq), all more recent developments that have required more hardware capability (and sometimes still employing samples as well), and all questionable as to whether they can come as close to an authentic simulation as a high quality sample set can (or as close to a real organ as a modeled organ can). As a rule, it looks like so far piano modeling requires more processing power but less RAM and/or storage than sampling.

Yamaha was a pioneer in modeled instrument sounds... see VL1, from 1993, which the net tells me was about $4k (adjusted for inflation, the equivalent of about $7k today). It had polyphony of two notes and focus on wind/reed instruments. Which might say something about the hardware required, since at about the same time, Yamaha was also selling the sample-based (and far more sonically diverse) SY85 for about half the price (with a polyphony of 30 notes). It would be really interesting to see Yamaha revisit that modeling tech today, with the much more affordable processing power available now. I guess it's a matter of whether Yamaha thinks there's much of a market for it.

Here's an interesting take on Yamaha using modeling for piano, which it seems they briefly did integrate into their sampled pianos in the CP1/4/40 series for the purpose of getting a better sounding piano out of less memory: http://sandsoftwaresound.net/spectral-c ... -modeling/

[Saul]

Pianoteq far exceeds the quality of any sampled piano I have ever come across and runs just fine on my ageing mid-2010 iMac running an Intel Core i3 3.2 GHz cpu with 8GB 1333 MHz DDR3 memory. Given the other things my mac is also having to run at the same time I don't think it is a stretch to say that modelling at this quality does not require uber amounts of processing power or ram.

Pianoteq also runs fine on my equally old PC.

Modelling is far more efficient than samples and more realistic.

[anotherscott]

Over at pianoworld, it's a commonly contested perspective, so it's clearly subjective... some agree with you that Pianoteq sounds better than sampled pianos, others vehemently disagree. Though I think there is virtually unanimous agreement that it *plays* very well, which is a different criteria. Anyway, back when it was developed, IIRC, pianoteq did require more processing power than sampled pianos... but the good tradeoff was that it required less memory. At any rate, nobody else can do piano modeling exacly the way pianoteq does it, because they have patents. That's the other thing about modeling being a generic term for using algorithms to duplicate sonic behaviors. Sampling is sampling, and at least broadly speaking, no company has different sampling capabilities than any other. But modeling is more of a concept, and every implementation can potentially be unique. Which also make it harder to generalize about what an implementation's hardware requirements might be.

[boothepr]

Its 2024 and I am having the same issue.

Whenever I layer piano, strings and pad, while using sustain pedal, playing all at the same time, the piano cuts out. I am very frustrated.

I have a MODX8+ and purchase it because of the increase in polyphony similar to Montage, which i thought would had fixed this issue. The issue persist. I have tried factory reset and it doesn't work. Is there a firmware upgrade to fix this?

[mx49]

Whenever I layer piano, strings and pad, while using sustain pedal, playing all at the same time, the piano cuts out.

Consider sharing the performance and an audio recording that demonstrates the issue. If possible also the corresponding recording of the MIDI events. If you do that, then people will be able to tell you why it is happening, and what you can do, so that it does not happen.

[anotherscott]

Its 2024 and I am having the same issue.

Whenever I layer piano, strings and pad, while using sustain pedal, playing all at the same time, the piano cuts out. I am very frustrated.

I have a MODX8+ and purchase it because of the increase in polyphony similar to Montage, which i thought would had fixed this issue. The issue persist. I have tried factory reset and it doesn't work. Is there a firmware upgrade to fix this?

There's almost certainly nothing to fix, it's probably working as designed:

If you're layering 3 AWM2 sounds, 128 note polyphony turns into 42 note polyphony (128 total divided by the 3 sounds you're playing)... and if any of your sounds, themselves, trigger multiple elements simultaneously (or multiple parts, if any of those sounds are, themselves, multi-part sounds), the polyphony will drop further. One way to address that is to select some different sounds. So if you're using an AWM2 string or pad sound, maybe you can find an FM-X sound that will work as well, which will then use the additionally available FM-X polyphony, and make more AWM2 polyphony available for the remaining sound(s). Or if you're using a sound that, itself, triggers multiple elements (or parts) per key, select a different one that doesn't. (For example, a piano with key-off noises will use polyphony more quickly than one without.)

Also, try being more judicious with your use of sustain pedal. Since piano sounds fade and string/pad sounds generally don't, I think sustain pedal on those other sounds can quickly get messy sounding anyway, so I'd only enable sustain pedal use for the piano sound, and turn it off for the string and pad sounds. If it sounds like those are then ending too abruptly, you can program a longer natural decay for them, which will give you a softer release without the potentially long holds of a sustain pedal.

And just in general, be aware of not holding the sustain pedal down longer than necessary. It could conceivably be an issue with your pedaling technique as well.