I should mention that not every note on a piano has dampers... it varies but notes up to G5 will have dampers and notes G#5 and above are undampered. They ring free at all times and are unaffected by Sustain pedal.

Yes, that much I knew & it's quite apparent when checking where the Key Off notes sound out and where they don't.

The Sustain pedal on a piano holds the dampers from returning to rest on the string. The dampers of a piano lift and fall for every note. Every note. The Sustain pedal simply prevents the dampers of those strings struck from falling back to rest. Even on keys struck with no Sustain pedal involved.

At rest the strings are dampened (the felts rest against the string preventing free vibration). The act of pressing a key causes a myriad of mechanisms to both lift the damper and strike the string. No Sustain pedal the felt damper returns to deadening the string as soon as you let go of the key. If the Sustain pedal is involved it simply holds this dampers up, allowing those strings to freely vibrate, until you finally let go of the pedal.

Run, don’t walk, to an acoustic piano. Play it and watch, listen... essential before attempting to emulate.

If my understanding is now correct regarding the mechanics on an actual acoustic grand, the dampers that lift & fall for every note would fall at the same loudness/velocity for each note, regardless of how hard or soft the keys were hit to begin with, correct..!? (Basically the weight of the damper falling back to the string would fall from the same height & be the same loudness every time whether a note was hit softly or really hard)

If this is true, then the "Key Off" noise for pianos should be set with a constant level & the velocity sensitivity should be removed, so that the volume is the same each time for each note regardless of what velocity the notes are hit with!

After further review having watched this video, where he compares the note ending damper sounds between a Steinway compared to a Fazioli (also a Bosendorfer), showing how different they are, it appears that the Key Off noise is dependent on how hard the note is hit, as well as soon or long after when the dampers fall, so I am now thinking that setting a constant level & removing the velocity sensitivity is not the best way to emulate the damper sounds...
https://youtu.be/mrB8IthgoRQ?t=1428 (Steinway)
https://youtu.be/mrB8IthgoRQ?t=1630 (Fazioli)

So Jason's envelope follower setting he described above is likely the best way at this point to make this the most realistic!

You can also hear that the lower down in the bass notes that you play, the more harmonic overtones that the Key Off damper notes have...I didn't listen to the CFX Key Off element waveform that closely regarding the lower notes vs mid or high notes, but I'll definitely be checking it out tonight...

I was able to get envelope follower to act as compression so the key off is louder when the strings are more resonant and not as loud when the instrument is essentially silent. You have to adjust the envelope follower gain to get a more "full swing" output out of it. This is often the case for any envelope follower work. The correct curve is a reverse ramp with negative polarity. Reverse ramps do not exist in the normal set - so you can either user define one or use "close enough" the "dogleg" curve type with Param 1 and Param 2 as "0" and a negative ratio with absolute value small enough so the "plateau" is minimized - but also large enough so the range of values is as wide as possible. It would be better not to have the "plateau" section - which you can accomplish with user curves - but even with it there, the effect was working "correctly" (take that with a grain of salt - I think I'd have to volume scale across the keyboard range to make it "better" ).

I'd like to try this, but have a few questions:

When you say adjust the envelope follower gain, or you referring to selecting "Part Param > Volume" or "Element > Element Level" ?
Also, where do I find "close enough" the "dogleg" curve type? Is close enough a preset Performance or PART? Does it have a curve defined & labelled as "dogleg"?
What exact settings do you use for the 'Absolute Value' ?

Envelope follower as in this:

https://www.yamahasynth.com/montage-category/envelope-follower

I followed part 1 although should follow a combo of 1 and 2. Works for everything below extreme velocities. Break down CFX Concert so you understand what the PARTs represent. Going by memory (not in front of keyboard) PART 1 is ppp to f all but highest non dampened strings. PART 2 is f to fff same range. PART 3 is high notes without dampening. PART 4 is only key off noise.

PART 4 you edit common PART area Mod/Control -> Control Assign. Dogleg is a built in curve type. I spelled out all but ratio because ratio is to taste. Experiment. It will be some negative value. Destination is level.

When you touch upper right to edit envelope follower 1, there is a new screen to adjust attack, decay, and gain. Crank the gain all the way up. If you're not editing this part of the follower, your results may seem like nothing is happening.

Envelope follower as in this:
PART 4 you edit common PART area Mod/Control -> Control Assign. Dogleg is a built in curve type. I spelled out all but ratio because ratio is to taste. Experiment. It will be some negative value. Destination is level.

So, I did a lot of experimenting with this last night. I presumed that you were setting the Polarity to 'Bi' and not 'Uni'; however I tried both and 'Bi' seemed to work the best of the two.
There is good news and bad news to report.

First the good news. It works! If I use the dogleg curve with the settings you listed above and set the ratio somewhere around -20 or less, there is a noticeable difference in volume after a period of time, whereby when releasing the sustain pedal the Key Off damper sound is much quieter than if you release it right away.

The bad news. As soon as you start setting the ratio to -1 or lower, the Key Off damper element notes start increasingly losing their release time, so that by the time you have the ratio set to something that works for the main problem, a great deal of the damper overtones are cut short and don't give the effect that they used to, especially for notes played without the sustain pedal. There is a noticeable difference and the damper overtones get a bit lost & release too soon for the effect it's supposed to give. Also when you do use the sustain pedal and hold it for several seconds, while it does work and the Key Off damper noise is at a more realistic volume, it is also too short in sustain when you do hear it, making it less realistic with not enough sustained overtones. Basically chokes the original sound off too soon.

I tried creating a custom User curve and it worked slightly better than the dogs leg curve, but it still took away the sustain of the damper sound/overtones.

When you touch upper right to edit envelope follower 1, there is a new screen to adjust attack, decay, and gain. Crank the gain all the way up. If you're not editing this part of the follower, your results may seem like nothing is happening.

This was interesting. As I moved the Gain from 0 to 24 I could see the led lights changing for the Input & Output, but the volume remained the same. It appeared to be doing something from what I could visually see, but there was no audible difference. (unless the audible difference would only be noticed when holding the sustain pedal for some period of time & then it changes the amount of reduction maybe..!?)

Almost! Unless there is a special custom User curve settings that I can use that would make this work as it did, but without losing the damper overtone sustain, this doesn't quite do it. It's unfortunate, because using envelope this way does fix the issue with releasing sustained notes without having the damper sound being too loud, but it just takes too much away from the damper sound itself... 🙁

I think the pianos sound great! They capture to me what im trying to play a piano. good job

So I haven't given up on making the Key Off Damper sound better and have just thought of one last thing to try.

Because the Envelope method does work, but in doing so it negatively affects the tail volume & length of the Key Off Damper waveform, I am going to try using 2 x Key Off Damper Elements, both of which will be set the same via the Envelope (element level, dogleg curve, negative absolute value ratio, parameters 0, 0); however for the second Element I am going to set the element to Delay before sounding (either by Key On Delay - Length or an AEG setting), set the AEG attack to slower so that it fades in, with goal of trying to blending in and bringing up the volume, as well as extending the length of the sound to be the same as Key Off Damper Element that has no Envelope configured.

Hopefully this works... 😉

What I think you're describing is that when you lift your hand off the keys - and the string sound quickly goes quiet - having the envelope follower squelch the key-off noise will over-squelch even when the strings are "loud" when you lift the keys because even though the envelope is "loud" at first - it will quickly go "soft" when you lift the keys and truncate the key-off noise.

There's a setting for this.

Remember that screen where you changed gain -- and didn't notice any difference? By the way - I'm not sure you're comprehending what you're changing - so I'm not sure you're thinking of this correctly to get a handle on how this is beneficial. I'll just skip past that for now. At any rate - this same screen has attack and release hysteresis (it's called something else in the GUI - maybe just "Attack" and "Release" ). Release has 10ms-680mS and attack 1mS-40mS. If you crank release up to 680mS - then this will "delay" the tail end of the envelope follower by nearly a second. This should be the first line of correcting the issue you're describing. 680mS isn't really much (just above 1/2 a second) - but maybe it will work. The key-off noise doesn't last for that long either.

Another way to fix this and still use envelope follower is to add an entire new PART ("burn" a PART slot) and send its output to OFF (not Main L&R). Even with output as OFF - the envelope follower still receives these PARTs. Then you can adjust the release envelope (AEG) of this PART so the PART continues to sound for a bit after you release the key. Sounding at a high enough volume that the key off noises are not overly squelched. With an entire PART dedicated to only producing an envelope for key off noise squelching - there's a lot of adjustments you can make to get this method to hit the intended target. Lots of flexibility. You can even use key-off noises on this PART that are actual loud tones that you tune the length of so you don't have to rely on release AEG if you don't want to. Remember that this added PART would need to velocity scale like the piano string PART. The other benefit to this is that the added PART would cover all velocities 0-127. The CFX Concert Envelope Follower 1 did not cover high velocities since PART 1 of that preset only covers lower velocity inputs.

... back to that gain setting.

Dogleg will create a curve like this:

When you have the gain set low, the envelope follower may only reach 64 at maximum (the mid-line on the x-axis of the curve). This means even though the followed PART is at its loudest - you're still applying a relative high negative offset to level. And so you're over-squelching when the input is loud. You really want zero level offset, if possible, when the input is following something loud. Setting the gain up makes the envelope's bar graph go higher to the right. Which is its output maximum. Which also means that when the output is loudest, the curve will go further to the right and apply less of a negative offset. Which is what you want. If you don't know what any of this is really doing, you'll be listening for the wrong thing. And you'll be stimulating the system the wrong way to determine what's going on.

Sooooo, how can i test this sample?

🙂

hello everyone. I have posted before about release samples. I have nord stage 3 and I have sampled nord royal 3d piano for montage, it is amazing for montage and the key point for great nord pianos (yamaha always overlooks in all keyboards ) is damper samples after key strike. it gives so realism to pianos and you can try and see with sample robot. Sample robot lets you record release sample after midi note out. If you cut this sections you can assign them to elements as key of sampling. I achieved this in yamaha montage.But the problem in yamaha this release sample's volume does not diminish if you continiously pressing the key. Nord keyboard diminishes the releases sample after 3-4 seconds of
continuous pressing of key. I tried envelope follower and motion sequencer for each key but I could't. Pressing damper pedal a spesific time and relasing it gives an unpleasant (high volume) damper noise. Any advice??

What I think you're describing is that when you lift your hand off the keys - and the string sound quickly goes quiet - having the envelope follower squelch the key-off noise will over-squelch even when the strings are "loud" when you lift the keys because even though the envelope is "loud" at first - it will quickly go "soft" when you lift the keys and truncate the key-off noise.

Yes, exactly!

There's a setting for this.
Remember that screen where you changed gain -- and didn't notice any difference? this same screen has attack and release hysteresis (it's called something else in the GUI - maybe just "Attack" and "Release" ). Release has 10ms-680mS and attack 1mS-40mS. If you crank release up to 680mS - then this will "delay" the tail end of the envelope follower by nearly a second. This should be the first line of correcting the issue you're describing. 680mS isn't really much (just above 1/2 a second) - but maybe it will work. The key-off noise doesn't last for that long either.

Ok, so setting it to 680 made a slight difference, but not enough.

Another way to fix this and still use envelope follower is to add an entire new PART ("burn" a PART slot) and send its output to OFF (not Main L&R). Even with output as OFF - the envelope follower still receives these PARTs. Then you can adjust the release envelope (AEG) of this PART so the PART continues to sound for a bit after you release the key. Sounding at a high enough volume that the key off noises are not overly squelched.

BINGO! I copied the CFX PART that has the ffff element, used just that element & setting the velocity from 0 to 127, then adjusting the release to 97. Fixed! I don't mind burning a PART for this because it's for just Piano 'Only' Performances. Any Performance with piano and strings, etc., I wouldn't care too much, but this works for me now! Thank You Jason!!!

... back to that gain setting.
When you have the gain set low, the envelope follower may only reach 64 at maximum (the mid-line on the x-axis of the curve). This means even though the followed PART is at its loudest - you're still applying a relative high negative offset to level. And so you're over-squelching when the input is loud. You really want zero level offset, if possible, when the input is following something loud. Setting the gain up makes the envelope's bar graph go higher to the right. Which is its output maximum. Which also means that when the output is loudest, the curve will go further to the right and apply less of a negative offset. Which is what you want. If you don't know what any of this is really doing, you'll be listening for the wrong thing. And you'll be stimulating the system the wrong way to determine what's going on.

You're right, I'm not fully comprehending what I'm changing; however your picture and explanation are great. I'll have to really read through it better to grasp it fully.
In a general sense, is it mainly affecting the sound level (how much negative volume is applied) after notes have been held for a period of time. So if set right, the Key Off release should be quieter with the Envelope Gain up higher, than if the Gain were just left at 0..!?

Although even with all this I still feel a fixed velocity is the best fix for key-off on CFX Concert (and others) - this is still as good as any reason to dig into what Envelope follower does, how it works, what unique tweaks can be made (gain, attack, decay on the follower), and how you can use it to get interesting results. Even if it's a lot to do for this particular use case (it'd be fair for some to shake their heads) - it's still worthwhile, if you're willing, to get something going with envelope follower even if you end up throwing the work away. What won't be thrown away is the experience gained and I often use envelope follower as a "get out of jail free" card when other methods do not work.

It's one of the only cross-PART modulation techniques where one PART can "talk" to another PART.

BINGO! I copied the CFX PART that has the ffff element, used just that element & setting the velocity from 0 to 127, then adjusting the release to 97. Fixed! I don't mind burning a PART for this because it's for just Piano 'Only' Performances. Any Performance with piano and strings, etc., I wouldn't care too much, but this works for me now! Thank You Jason!!!

You may have done this, but be sure this new burned velocity=0-127 PART has its output set to OFF instead of Main L&R to prevent from having phase cancelation type issues or additive issues (too loud combined) from two PARTs both triggered and playing the same sample.

And a minor point:
Since all you're doing is envelope - there's not really a need to pick a CFX sample as the envelope source. It would be "best" to pick a source sound (maybe even FM-X) that has the widest dynamic range and a response curve that suits what you're trying to do on the squelch side. Not all samples will react the same and certainly there's a lot of flexibility in programming of FM-X which is why I tend to use it as a stimulus vs. samples (AWM2). However, that may not be something you want to tackle which would be understandable if that's the case. But if there happens to be any single-PART AWM2 sound you think has great softs and great louds (wide range) - that'd be a good candidate. I haven't looked at presets this way - so I don't have a crystal ball here. Maybe CFX you've picked is close enough to being this -- maybe not.

As always - it's the results that matter. If it's working for you - then that's what counts.

In a general sense, is it mainly affecting the sound level (how much negative volume is applied) after notes have been held for a period of time. So if set right, the Key Off release should be quieter with the Envelope Gain up higher, than if the Gain were just left at 0..!?

Envelope follower provides a number between 0 and 127. This number represents how loud the PART(s) the Envelope Follower is "following" is being at the current moment (ignoring attack/decay extremes). So Envelope Follower will "listen" to the PART you designate (envelope follower 1 is for PART 1, ... envelope follower 8 is for PART 8, ... envelope followers 17 and 18 are for A/D inputs L and R). When the followed PART is silent, the envelope follower will output a 0. When the followed PART is saturating the output (loud) - the envelope follower will be 127. This assumes unity gain. Unity gain maining you are neither boosting nor trimming the envelope's output. It's as if there's a ghost in your keyboard who's job it is to turn a dial between 0 and 127. If the ghost hears nothing, he instantly turns to 0. If his ears are blown out, he dials it in to 127. As the sound is getting louder and softer, he dials up and down to follow the sound's amplitude (loudness). The ghost is able to hear the intended PART by holding a virtual microphone up to the PART's output. Turning up the gain just brings the virtual microphone closer to the source of the PART's output and turning down the gain brings the microphone further away. Not really, but a visualization that's close enough to the truth (there's not really a ghost in the machine).

So turning up or down the gain doesn't care about how long you've held a note. Envelope follower is going to follow what it hears as the PART's output at the current moment. There are some sounds that die down as you hold the note. Naturally, the envelope follower would decrease its value as the sound's volume fades. Some sounds can actually get louder as you hold a key down. In this case, the envelope follower would increase value as you hold down a key.

Gain just scales the value the envelope follower will output. Like bringing the microphone closer would do.

If you strike keys and notice the envelope follower bar graph pegs out at 127 - then you'd want to bring the gain down so it doesn't reach a saturation point and stop showing differences. Likewise, if you see the maximum value cannot ever reach 127 - this *may* be an argument to increase the gain. It really depends if you want full swing or not. You may not want full swing.

For this, I wanted to try to achieve full swing. That's "full" swing, really. It's difficult to get 127 out of the envelope follower since generally PARTs are intentionally programmed to have lots of ceiling space so PARTs can be combined and together not saturate the amplitude (peak). It's even difficult to do when you try to saturate with a PART that's turned its output off and would do no harm in saturating.

Hello Darryl and Jason. Today it is over and i can see the potencial of montage. I have modelled the nord royal grand 3d with sample robot with release samples whics reduces with time and without annoying damper noises. Forget about envelope follower. The key point is motion sequencer. At first tbe level of release samples and the piano samples should be same. (Etc. 100) then edit the release sample part. Enter the motion sequencer and select Lane 1. I prefer 8 cyle sequence with square type. The each sequence should increase after each other. The amplitude value should be 63. Increase tbe smoothness to maximum level. After then enter the control assign page and select lane one and part volume. The curve type should be logarytmic and type 3 and - 35 value (minus) the each sequence trigeer should be selected aftet each note and the sequencer loop should be selected of. It is over. Thanks.