Akoustik Piano (very short review) [Archive]

View Full Version : Akoustik Piano (very short review)

Bogdan

11-22-2005, 09:43 PM

Just got Akoustik Piano from NI. Installed 24-bit version. So far I tried mostly Steinway D and these are the first impressions:

10 velocity layers are really 5+5. I'm guessing 5 for pedal down and 5 for pedal up
very good quality piano was sampled (not just because they said so, but because it really sounds as it was in good shape)
EQ that is provided works good
it has a nice way to adjust resonances (both sustain and release)
good sounding convolution reverb with 4 different spaces
microphone position can be adjusted but it's just simulated, not real samples
dynamic range for the attack portion of samples was not what I expected based on the marketing information (only about 15-17dB range between loudest and softest samples for pedal up and about 8-9dB for pedal down)
only white keys were sampled
there is a good review in Keyboard magazine December 2005 issue (they didn't figure out that it has only 5 layers and that dynamic range is not that great as a consequence...)

In summary, this is very good piano that has only one problem. The dynamic range is too narrow. The pp or ppp velocities sound as mf with turned down volume (if pedal down) or p with turned down volume (if pedal up). The timbre of the soft notes is not satisfactory. So, if you play pp or ppp and switch pedal up and down as you go, it would sound bad.

However, it was very well tuned and with very solid mechanics.

If you use this piano within a recording that has more instruments, most of these little problems would be masked anyways.

If you use it as a solo piano, you should play only at or above mp or mf. The p and below will not be that good. So, it will not work that good for soft played Jazz trio or similar combos.

This library is worth having.

But, I'm still looking for one that would have all that this one has plus better dynamic range that is real, not simulated by turning volume up or down.

This piano is so far the closest to the real thing. It's a shame they did not do a bit more balanced sampling and did not provide more real velocity layers rather than only 5 loudest ones.

Bogdan

P.S. I'll create another thread regarding piano velocity layers in general to check what you guys think about some ideas that crossed my mind recently.

irvind

11-23-2005, 05:42 AM

Try Sampletekk's new TBO (The Big One). 31 sampled velocities per note pedal up, 31 velocities per note pedal down, and 31 release velocities per note. It's awesome. It's the (sampled) expressive piano I've been waiting for for years. pp to ff, no problem.

xav93

11-23-2005, 06:28 AM

10 velocity layers are really 5+5. I'm guessing 5 for pedal down and 5 for pedal up

Not right : it's 10 velocity layers for pedal up, 1 for pedal down, and 1 for release.
When I was working on the August Foerster in K2 sound library, I gave NI a soluce to use only 1 "pedal down" layer and however to sound real (a sampling soluce only for pedal down samples). They used this idea on the August Foerster and I assume they used the same idea in AKP.

Lougheed

11-23-2005, 06:38 AM

I also recall the video that introduced Akoustik, and the dude talked about some kind of processing to give many layers, some real and some virtual, I would assume.

How does it play?

Does it "feel" and respond like only 5-layers?

BTW, thanks for the review.

Lawrence

xav93

11-23-2005, 07:43 AM

fozzy

11-23-2005, 08:42 AM

This library is worth having.

Thanks for posting this review! It's helpful. The jury is still out on this one for me. After Bruce's comments on the TBO that will likely be the next purchase, then I'll think closely about this one.

Bogdan

11-23-2005, 09:09 AM

Sure it do not sound like a 5 layers... because it has 10 (see my previous post).

NI used K2 script to smooth transitions between layers, so it appears as if it has more than the 10 recorded. This script is a very good idea and is well programmed. It uses features from K2.0.8.002. from which Akoustik Player is issued.

Interesting. I listened to the wave files for various notes on Steinway D. They are grouped as 5+4+1. The last in a group is the loudest sample.

Now, I did see also additional resonance samples as well as something that could have been pedal down (single sample).

So, if you're right the problem might be that the timbre of the 10 (5+4+1) layers that are provided is not real good in terms of dynamic spread. The softest sample sounds more like p, not ppp. (timbre-wise) I tried processing samples and reducing volume to see if scaling was done on the two groups. It is possible. However, the layers are very close to each other in terms of timbre. I don't like that. Steinway doesn't sound like that at ppp. The recording process did not appear to be uniform as far as I could see. Maybe the issue was a choice of Mic or Mic placement, or the force they used for triggering the notes when sampling the sound.

I was listening to the raw wave files for the individual notes trying to find which ones would be ppp or pp samples. Didn't find any that sounded softer than p regarding the timbre, even after processing them.

So, when I played piano (a duet with Bass and Piano), I noticed that some low register notes do not sound right when played soft. That correlated with previous observations I made. (I tried velocity curves, etc. as one would do before starting serious testing or playing, I also monitored the actual ppp or pp assignments in their window - very nice feature!)

If they have 10 layers for pedal up it was done in such a way that you hear really only 4-6 or so since they are spaced very close in terms of timbre. When playing pedal up/down it doesn't sound good. (as far as I'm concerned) :)

Maybe the reason for a weird feel when playing this piano is that they really used only one layer for pedal down and they did crossfading into scaled pedal down immediately following the attack portion of the pedal up sample. Well, it didn't work well for my music. Most likely since the pedal down phase properties over the duration of a sample will not be proper for all velocity layers. ( a little bit of a Frankenstein type sound design, I guess :)

Next, the method they used for crossfading layers might be nice (in theory) and it does make it hard to hear where the switching occurs (if you just consider loudness). However, it makes the piano sound a bit artificial since the timbre change does not follow the loudness change. That is, they failed short from getting proper timbre changes as they go between the layers. (especially when you use pedal down) You can't get proper timbre change between layers by simply using linear combination of two layers (doesn't matter what curve you use, the operation is linear which is not what happens in a real piano)

They may have provided more layers above f toward the ff and fff (that 4+1 group could be that), with possibly 5 layers between f and p.

It is possible that since they use a single pedal down sample (with attack portion cut off?) they could not really use any softer pedal up samples than 'p'. The attack portion of softer pedal up samples may not "connect" as good to the single available pedal down sound that would follow?

We should keep in mind that if you play a pedal down note a few times in a row with different velocity, you do not want to hear the same timbre resonating out each time you play the note... This is possibly a single thing that kills the otherwise great performance of this piano!

Having said all this, I still think the piano is good. Don't get me wrong. It is well tuned (only one or two notes in the low and one or two in the high register may stick out a tiny bit). Mechanics was in a great shape as far as I could hear.

So, the piano is valuable and I will be able to use it. But, I may have to do a lot of EQ and tweaking to get it to sound right and smooth in solo pieces or small jazz combo sets.

Thanks for pointing it out that only one pedal down sample might be used here.

Some people may found this piano to be a cheat. When I see 10 layers advertised, I would like to get 10 for pedal up, 10 for pedal down as a minimum! It is as simple as that. I know from the experience that trying to do "traditional" processing (including advanced DSP) will not help you get a single pedal down layer to sound like 10 (unless you do some serious physical modeling of the piano acoustics :) )

Regards,
Bogdan

tony patterson

11-23-2005, 09:52 AM

Interesting. I listened to the wave files for various notes on Steinway D. They are grouped as 5+4+1. The last in a group is the loudest sample.

Now, I did see also additional resonance samples as well as something that could have been pedal down (single sample).

So, if you're right the problem might be that the timbre of the 10 (5+4+1) layers that are provided is not real good in terms of dynamic spread. The softest sample sounds more like p, not ppp. (timbre-wise) I tried processing samples and reducing volume to see if scaling was done on the two groups. It is possible. However, the layers are very close to each other in terms of timbre. I don't like that. Steinway doesn't sound like that at ppp. The recording process did not appear to be uniform as far as I could see. Maybe the issue was a choice of Mic or Mic placement, or the force they used for triggering the notes when sampling the sound.

I was listening to the raw wave files for the individual notes trying to find which ones would be ppp or pp samples. Didn't find any that sounded softer than p regarding the timbre, even after processing them.

So, when I played piano (a duet with Bass and Piano), I noticed that some low register notes do not sound right when played soft. That correlated with previous observations I made. (I tried velocity curves, etc. as one would do before starting serious testing or playing, I also monitored the actual ppp or pp assignments in their window - very nice feature!)

If they have 10 layers for pedal up it was done in such a way that you hear really only 4-6 or so since they are spaced very close in terms of timbre. When playing pedal up/down it doesn't sound good. (as far as I'm concerned) :)

Maybe the reason for a weird feel when playing this piano is that they really used only one layer for pedal down and they did crossfading into scaled pedal down immediately following the attack portion of the pedal up sample. Well, it didn't work well for my music. Most likely since the pedal down phase properties over the duration of a sample will not be proper for all velocity layers. ( a little bit of a Frankenstein type sound design, I guess :)

Next, the method they used for crossfading layers might be nice (in theory) and it does make it hard to hear where the switching occurs (if you just consider loudness). However, it makes the piano sound a bit artificial since the timbre change does not follow the loudness change. That is, they failed short from getting proper timbre changes as they go between the layers. (especially when you use pedal down) You can't get proper timbre change between layers by simply using linear combination of two layers (doesn't matter what curve you use, the operation is linear which is not what happens in a real piano)

They may have provided more layers above f toward the ff and fff (that 4+1 group could be that), with possibly 5 layers between f and p.

It is possible that since they use a single pedal down sample (with attack portion cut off?) they could not really use any softer pedal up samples than 'p'. The attack portion of softer pedal up samples may not "connect" as good to the single available pedal down sound that would follow?

We should keep in mind that if you play a pedal down note a few times in a row with different velocity, you do not want to hear the same timbre resonating out each time you play the note... This is possibly a single thing that kills the otherwise great performance of this piano!

Having said all this, I still think the piano is good. Don't get me wrong. It is well tuned (only one or two notes in the low and one or two in the high register may stick out a tiny bit). Mechanics was in a great shape as far as I could hear.

So, the piano is valuable and I will be able to use it. But, I may have to do a lot of EQ and tweaking to get it to sound right and smooth in solo pieces or small jazz combo sets.

Thanks for pointing it out that only one pedal down sample might be used here.

Some people may found this piano to be a cheat. When I see 10 layers advertised, I would like to get 10 for pedal up, 10 for pedal down as a minimum! It is as simple as that. I know from the experience that trying to do "traditional" processing (including advanced DSP) will not help you get a single pedal down layer to sound like 10 (unless you do some serious physical modeling of the piano acoustics :) )

Regards,
Bogdan

To be honest, I favour the Bechstein over the Steinway!
Tony

Keith Jarrett

11-23-2005, 10:13 AM

If you use it as a solo piano, you should play only at or above mp or mf. The p and below will not be that good. So, it will not work that good for soft played Jazz trio or similar combos.

I'm not agree with that http://www.northernsounds.com/forum/images/smilies/biggrin.gif

This is a personal record with the AP Steinway D: Over The Rainbow (http://jfbcda.club.fr/Over%20The%20Rainbow.mp3)

xav93

11-23-2005, 10:25 AM

Olivier

11-23-2005, 11:10 AM

howardv

11-23-2005, 12:02 PM

So they're using some kind of dynamic filtering scheme applied to a single layer of pedal down samples? As oposed to deriving pedal down sounds by convolving the pedal up layers? Most libraries give you a full-depth pedal down layer as an option.

Howard

Bogdan

11-23-2005, 01:09 PM

I'm not agree with that http://www.northernsounds.com/forum/images/smilies/biggrin.gif

This is a personal record with the AP Steinway D: Over The Rainbow (http://jfbcda.club.fr/Over%20The%20Rainbow.mp3)

Sounds good.

Would you be willing to share the preset settings you used, type of controller keyboard (unless it was a piano roll), etc. You can e-mail it to me using private message if you are not willing to share with others.

I would really like to get this kind of response and tonal quality, but it escapes me at the time being.... :confused:

Now, after I listened the example for the second time, I did hear that low register often sounds as just a turned down volume... :confused: On some notes and in some situations.

Overall, I got feeling after listening 3 times in a row that the tonal quality is a little bit razor sharp even for the soft notes....

To repeat. I like this piano. It's not I don't like it. I just think it is a shame it did not provide better selection of velocity layers that would result in more timbral variety as it would be on a real piano.

Regards,
Bogdan

xav93

11-23-2005, 01:31 PM

So they're using some kind of dynamic filtering scheme applied to a single layer of pedal down samples?
No. That's not it.

Bogdan

11-23-2005, 01:32 PM

All you quote here is for a "traditionnal" mapping and programmation.

When I talk about the only pedal down layer, you think that it has been traditionnaly recorded and traditionnaly programmed. That's far not the case. I don't think you know how it has been recorded (and how it's simple to do it), but it makes difference with all classic sampling. I can give you a clue : maybe in analysing only the pedal down samples with a spectral analyser you will understand... But I don't want to tell all making's secret...
The timbre when you play pedal down is far to be the same as you believe. Because I said there is only one pedal down layer, you imagine false things. It seems your ears betray you.
You speak about crossfading : who talked about crossfade? You only suppose and that's not the case.

But I agree with you about the fact there are not enough layers in ppp.

I'll take a look when I get home today...

Since you know so much about it, could you just tell me one thing: If you play a single note with pedal down at different velocity, would you get the same/similar timbre with the effect like turning down the volume? (yes/no - you don't need to disclose any details)

Reason I'm asking is that I tried for example note F1/F#1 and had very bad luck trying to get it to sound real soft with pedal down and with long enough decay time.

Maybe all of this is a consequence of not having enough granularity on ppp side?

BTW, I listened to "pedal down sample" and it did not really sound like a piano note (you're right, next I started assuming certain things, but I'll get to my DSP toolkit later today since you guys prompted me to do more analysis :) )

I guess, I'm just trying to get this piano to sound as good and real as possible and so far I did not have much luck with soft notes...

Which convolution filter would you use (out of those provided) to get better soft note qualities? Any hints on preset settings that could be helpful (other than just velocity curve selections or EQ)?

Thanks for the information so far!

BTW, providing so easy access to EQ in this piano is a very nice idea... (although I'd like to see velocity sensitivity being configurable as well so that the EQ could react on velocity instead of being fixed as I had to suppose again :) )

Regards,
Bogdan

xav93

11-23-2005, 03:46 PM

I'll take a look when I get home today...

Since you know so much about it, could you just tell me one thing: If you play a single note with pedal down at different velocity, would you get the same/similar timbre with the effect like turning down the volume? (yes/no - you don't need to disclose any details)
Pedal down resonance behaviour, as you may know, is a bit more complex than only following midi velocity and his linked volume. While sustain pedal is down, when you hit many times the same key at different velocities, it's always the loudest that keeps on sounding.
That's not the same behaviour for the strings : each string is actually monophonic. So simulating the transitions between a strong hit to a soft hit can be hard. About this (but not only this), a friend of mine, Olivier Frappier, sent to NI a nki patch he has improved.

Reason I'm asking is that I tried for example note F1/F#1 and had very bad luck trying to get it to sound real soft with pedal down and with long enough decay time.
I've just tried it and I have not such thing.

Maybe all of this is a consequence of not having enough granularity on ppp side?
I don't know if this will help you, but I'm used to set Akoustik Piano's velocity curve to -2 and my master keyboard (a RD700sx) response (not velocity curve, but "key touch") to heavy-5, and I find soft playing quite good. To set correctly the master keyboard response is very crucial for best result.
Sure, timbre is far not as warm as than TBO's ppp . ppp layer is maybe the weak point of this AKP Steinway D. But it's hard to get full timbre range with "only" 10 layers.
I've just tried also this : AKP velocity curve set to -3 and, because of too low volume while soft playing, set dB dynamic to -1 (or -2). I think it's better.

Bogdan

11-23-2005, 04:45 PM

Pedal down resonance behaviour, as you may know, is a bit more complex than only following midi velocity and his linked volume. While sustain pedal is down, when you hit many times the same key at different velocities, it's always the loudest that keeps on sounding.Sorry, but not true on a real piano. Actually, if a sampler cannot play multiple same notes it is a lousy sampler as far as I'm concerned. Gigastudio does it very well as far as I can tell...

I don't know if this will help you, but I'm used to set Akoustik Piano's velocity curve to -2 and my master keyboard (a RD700sx) response (not velocity curve, but "key touch") to heavy-5, and I find soft playing quite good. To set correctly the master keyboard response is very crucial for best result.
Sure, timbre is far not as warm as than TBO's ppp . ppp layer is maybe the weak point of this AKP Steinway D. But it's hard to get full timbre range with "only" 10 layers.
I've just tried also this : AKP velocity curve set to -3 and, because of too low volume while soft playing, set dB dynamic to -1 (or -2). I think it's better.Thanks for the suggestions. I'm in front of it now and I'll be trying it out. I arrived to the same velocity curve as well.

I did a few things today before reading this message. I tried again the low register and I'm now convinced that the low register has more problems than the mid and high notes. The low register notes octave #1 for example. C1-B1 notes don't sound well (soft played). I watched the "Last Note" display and could see when it is ppp or pp or p. The p layer definitely sounds more like mp or even mf. The pp/ppp sound essentially the same, and the timbre is not too good.

Nevertheless, I'll keep playing with it. Everything else with this piano is simply great. It makes me smile when I hear how well it is tuned and how well it responds in all other ranges but perhaps this one or one and a half low octaves.

Best Regards and Thank you for your suggestions,
Bogdan

xav93

11-23-2005, 05:57 PM

Originally Posted by xav93
Pedal down resonance behaviour, as you may know, is a bit more complex than only following midi velocity and his linked volume. While sustain pedal is down, when you hit many times the same key at different velocities, it's always the loudest that keeps on sounding.

Sorry, but not true on a real piano. Actually, if a sampler cannot play multiple same notes it is a lousy sampler as far as I'm concerned. Gigastudio does it very well as far as I can tell...

I wasn't as clear as usual in my explaination. I'm only talking about resonance, not sample. Try on a real piano : while sustain pedal is kept down, play C3 hard then C3 soft. "C3 soft" is obviously sounding, but with "C3 hard" resonance. When C3's string (3 strings in reality) is softly hit after the hard hit, it becomes to be less vibrating (a string has a monophonic behaviour). But as other strings (of all other keys) are undampered because sustain pedal is still down, the resonance is unchanged.

In the case where the hard note follows the soft note, it's different. Soft note and soft resonance are masked by hard note and hard resonance.

Bogdan

11-23-2005, 10:50 PM

I wasn't as clear as usual in my explaination. I'm only talking about resonance, not sample. Try on a real piano : while sustain pedal is kept down, play C3 hard then C3 soft. "C3 soft" is obviously sounding, but with "C3 hard" resonance. When C3's string (3 strings in reality) is softly hit after the hard hit, it becomes to be less vibrating (a string has a monophonic behaviour). But as other strings (of all other keys) are undampered because sustain pedal is still down, the resonance is unchanged.

In the case where the hard note follows the soft note, it's different. Soft note and soft resonance are masked by hard note and hard resonance.That is correct.

I understand. (sometimes it is hard to be clear using text messages, happens to me all the time :) )

I just tried this piano a bit more using Una Corda and Sostenuto pedals. Works real nice. As I was playing using Una Corda pedal I recalled the analysis I did on wave files form the Steinway D. It came to me that the two groups of samples could be representing notes played with or without the Una Corda pedal.

So, I did a bit more listening tests and analysis and determined that this is the most likely case.

So, I think I solved the mystery of Akoustik Piano layers. (all I can do is suppose again since I don't know for a fact what they really did) Nevertheless, an educated guess would be as follows:

5 layers for Una Corda Pedal Down (the softer sounding samples)
4+1 layer for Una Corda Pedal Up (the brighter sounding samples)

The reason I believe this is two-fold:

1. The spectral properties of two groups are consistent with Una Corda Pedal being used on a first group (has the faster roll off toward high freqs as you would expect when Una Corda is used, i.e. creates a low pass filtering effect)

2. The signal envelopes (in time domain) for the two groups match for comparable average peak powers. One for Una Corda Down and one for Una Corda Up.

So, I think what they did was exactly that. They sampled 5 layers for Una Corda Down and 5 for Una Corda Up.

They did a fancy approach to Sustain Pedal handling as you mentioned. (I'm still looking into that as a nice puzzle) It is nice idea. I will just have to think about it for some time and look into pros and cons form the physical modeling prospective...

Great talking with you!

Regards,
Bogdan

Bogdan

11-23-2005, 11:23 PM

[...] When C3's string (3 strings in reality) is softly hit after the hard hit, it becomes to be less vibrating (a string has a monophonic behaviour). But as other strings (of all other keys) are undampered because sustain pedal is still down, the resonance is unchanged.

In the case where the hard note follows the soft note, it's different. Soft note and soft resonance are masked by hard note and hard resonance.I thought about this a bit more...

There might be two problems with this approach:

1. When you hit a key hard compared to when you hit it less hard you may not get the same type of resonance (you may encounter some nonlinear acoustic effects from the soundboard as well as frequency dependent changes in the timbre from the resonating strings)

2. As you keep repeating the note while the sustain pedal is down, and keep changing the velocity while you are doing it, the timbre you would build up would depend greatly on when do you hit subsequent notes relative to the decay of the resonance. Since the resonances are not linearly related on a real piano you would be doing just a first order approximation of the sound for the pedal down.

Now, we come to very interesting point. This problem is one that cannot be solved by samplers! Period. (In my opinion) You would have to do physical modeling combined with sampling to get close to the real thing. Problem with using pure pedal down samples here is that the subsequent notes on a sampler add up linearly and do not interact with each other as they would on a real piano that has particular soundboard. Samplers freeze the state of a soundboard when a single note is played. That's what they record. Once you start playing repeated notes, samplers just keep linearly adding samples up (scale and add, scale and add, and so on). Real piano doesn't work that way. The soundboard and the body of a piano would introduce their own response which may not be linear...

Very interesting point. I think this might be one of the issues that samplers have a problem with when you get to sampling of acoustic instruments. Samplers in general can sum things linearly. They are not aware of any nonlinear interactions between the subsequent notes on various acoustic instruments including the piano.

[Again, this is a guess. I'm not 100% sure, but since I played real piano for so long I know for a fact that I can generate very interesting sounds with pedal down and repeated notes of varying loudness... I was never able to do the same on any sampler so far.]

Regards,
Bogdan

Olivier

11-24-2005, 03:33 AM

I thought about this a bit more...

There might be two problems with this approach:

1. When you hit a key hard compared to when you hit it less hard you may not get the same type of resonance (you may encounter some nonlinear acoustic effects from the soundboard as well as frequency dependent changes in the timbre from the resonating strings)

2. As you keep repeating the note while the sustain pedal is down, and keep changing the velocity while you are doing it, the timbre you would build up would depend greatly on when do you hit subsequent notes relative to the decay of the resonance. Since the resonances are not linearly related on a real piano you would be doing just a first order approximation of the sound for the pedal down.

Now, we come to very interesting point. This problem is one that cannot be solved by samplers! Period. (In my opinion) You would have to do physical modeling combined with sampling to get close to the real thing. Problem with using pure pedal down samples here is that the subsequent notes on a sampler add up linearly and do not interact with each other as they would on a real piano that has particular soundboard. Samplers freeze the state of a soundboard when a single note is played. That's what they record. Once you start playing repeated notes, samplers just keep linearly adding samples up (scale and add, scale and add, and so on). Real piano doesn't work that way. The soundboard and the body of a piano would introduce their own response which may not be linear...

Very interesting point. I think this might be one of the issues that samplers have a problem with when you get to sampling of acoustic instruments. Samplers in general can sum things linearly. They are not aware of any nonlinear interactions between the subsequent notes on various acoustic instruments including the piano.

[Again, this is a guess. I'm not 100% sure, but since I played real piano for so long I know for a fact that I can generate very interesting sounds with pedal down and repeated notes of varying loudness... I was never able to do the same on any sampler so far.]

Regards,
Bogdan

In NI programming you haven't a linear self-mask.

it's depend on velocity and on size of the samples, on the time, on the attack, on the decay and others parameters.

Regards,

Olivier

xav93

11-24-2005, 03:41 AM

1. When you hit a key hard compared to when you hit it less hard you may not get the same type of resonance (you may encounter some nonlinear acoustic effects from the soundboard as well as frequency dependent changes in the timbre from the resonating strings)

Agree. And then I would become interessant to use in addition or not (I don't know yet) with some pedal down samples, some kinds of dynamic convolution, with many piano's impulses. I learned yesterday that dynamic convolution processors already exists.

2. As you keep repeating the note while the sustain pedal is down, and keep changing the velocity while you are doing it, the timbre you would build up would depend greatly on when do you hit subsequent notes relative to the decay of the resonance. Since the resonances are not linearly related on a real piano you would be doing just a first order approximation of the sound for the pedal down.
Agree, but NI as been clever on this. I don't tell more.

Now, we come to very interesting point. This problem is one that cannot be solved by samplers! Period. (In my opinion) You would have to do physical modeling combined with sampling to get close to the real thing. Problem with using pure pedal down samples here is that the subsequent notes on a sampler add up linearly and do not interact with each other as they would on a real piano that has particular soundboard. Samplers freeze the state of a soundboard when a single note is played. That's what they record. Once you start playing repeated notes, samplers just keep linearly adding samples up (scale and add, scale and add, and so on). Real piano doesn't work that way. The soundboard and the body of a piano would introduce their own response which may not be linear...
About period, notice I don't know if you speak of the period of a periodic phenomenon or perid only as time, but for time related phenomenon, NI made interesting things in AKP script also
Again, I think this could be partially solved with dynamic convolution. Results look good with only one impulse (pedal up and/or pedal down). But sure, non linear phenomenons could be better emulated by using dynamic convolution (pedal up and/or pedal down). But this could be able by using K2.0.8.002 script. You can program in it non-linear rules that affect so much things (as selecting an impulse file or controlling correctly "send FX" to different convolution FX...)