Piano pedaling realism

[Bruce A. Richardson]

Hi Steve,

Post Grandioso has a pedal emulation mapping that does indeed switch the samples in mid-note.

However, it\'s very crude, as one would expect. This is really much more in the realm of a physical modeling hybrid than a sampler, and we\'re going to be waiting a few more hardware and software generations to start seeing that kind of computational ability in a realtime softsynth.

I really like the Post library, but I wouldn\'t rate the pedaling emulation as one of the strong points. In a limited and controlled application, it can probably be made to sound very real, but overall, I have not had that experience with it.

[midphase]

I suppose that by sending a controller 64 to do a quick xfade between different sample layers it would be possible to emulate the effect that you\'re after. It would be tricky to devise a curve that would be smooth enough to not cause any clicking or popping, but quick enough to avoid any phasing or a 6db amplitude peak.

Of course you\'d also be halfing the polyphony, if you add release triggers and a few other bells and whistles, a GS 160 will be reduced to a GS 32 in no time!

I think it\'s doable.....is it worth it? Perhaps, but except for solo piano situations, I think the effect would be too subtle to justify using it in busier mixes!

[Jeff Hurchalla]

This is something I\'ll be doing for Michiel, and should work much like Midphase describes, think of it as a triggered crossfade initiated with the sustain pedal, though in practice phasing wasn\'t really a problem. And Bruce, I think the original intenet with the Grandioso patches you loaded was for use with the crossfade, so it\'s not too surprising that it seems like it\'s not getting the repedal effect quite right. Hopefully it will get this sound across just right, or at least so that it sounds satisfying for folks. Yep polyphony is an issue, but depending what you\'re looking for I think it can be justified. I like the option, and as the technology advances polyphony should be getting to be less of a problem.

[pantonality]

Well midphase guessed it. I\'m working on a solo piano piece and there are things that are common practice that just don\'t work. It\'s also a PITA from a performance perspective. I always play the notes then push down the pedal, that\'s just the way an acoustic piano sounds best. I undertasnd the difficulties involved in programming it, but I also figured it would be better if a programmer programmed than if I tried my hand at it. Based on the sample I heard from Michiel Post a few weeks back and Jeff\'s indication that he was working on this very problem for their upcoming release that for me will seems to be a piano library worth waiting for. So Jeff hurry up and get back to work, stop browsing the net, images/icons/grin.gif ...and I look forward to seeing that library available soon.

Steve

[Bruce A. Richardson]

Originally posted by Jeff Hurchalla:
This is something I\'ll be doing for Michiel, and should work much like Midphase describes, think of it as a triggered crossfade initiated with the sustain pedal, though in practice phasing wasn\'t really a problem. And Bruce, I think the original intenet with the Grandioso patches you loaded was for use with the crossfade, so it\'s not too surprising that it seems like it\'s not getting the repedal effect quite right.<font size=\"2\" face=\"Verdana, Arial\">That would explain it, for sure. It was a hard switch.

[eliam]

I\'ve been working with the Imperial grand Bosendorfer and I find that the pedal down dimension isn\'t truly realistic. I want to be able to control my mix of note/general resonance, and that is why I use resonance-only samples, which were made like that: as soon as the hammer hit the string, the same string was muffled with the finger and thus only the resonance of the whole instrument was recorded. This works fine, and I only have to choose which notes will trigger the damped samples (because if I add them to every pedal down note, it becomes too charged).
I believe that the pedal down samples do not permit enough control, and -for what I\'ve heard- do not produce a realistic effect. The damped string samples might be a solution to this...

[pantonality]

I have Gigapiano and find that the response to the sustain pedal is unrealistic. If the pedal is down when a note is sounded the resonance sample gets used. If you push the pedal down after you play the note it has no effect. Of course on a real piano the sound shifts every time the pedal is used. Do any of the other piano libraries emulate the response of the sustain pedal on a piano. Can you play a large chord with the pedal down then let the pedal up and get the nonresonance sound?

Steve Chandler
http://www.mp3.com/stevechandler (\"http://www.mp3.com/stevechandler\")

[Jan]

...And then there is half pedaling, quarter pedaling, and they affect the colour of the sound of the piano to such an extent that mere sampling will not suffice.

[eliam]

How\'s that? I\'m not familiar with half-pedals...

[pantonality]

Hi Eliam,

Half pedals are when you depress the sustain pedal and let it up, but not all the way, the depress it again. Jan is correct in his assertion that it would be difficult to perfectly emulate the bahavior of a pedal with samples. I understand that, but that\'s not what I asked for. The responses seem to indicate that what I\'ve asked for is being worked on by at least Michiel Post for an upcoming library release. Is anyone else working on this?

Steve

[Bruce A. Richardson]

Yes, the thing about the pedaling issue is that there are a thousand shades of gray. The resonances change depending on the chord struck, the way the pedal is pressed...there is really a great deal of difficulty here in successfully modeling the behavior of a piano when you factor every complicating scenario.

Frankly, I\'ll have to hear a successful \"crossfade\" before I believe it will be a truly valid mapping technique--that is before I accept that the pros will outweigh the cons. So far, I don\'t...going between no resonance to full resonance and back isn\'t likely to sound natural no matter how well the crossfade is finessed, and you\'ll still have that uncomfortable few milliseconds where phase relationships will conflict. There\'s no way to phase lock samples of piano length for the entire length of a sustain. At some point, you\'re going to get the artifacts associated with two strings.

I think the idea of recording \"just\" resonance, while damping the struck note\'s string is a good one for this scenario...it\'s an additive methodology, and will have its own implementation challenges, but it does solve the problem. Certainly worth experimenting.

[Jake Johnson]

Any advantage to the Maj method of recording only pedal down samples, so it\'s a matter or reducing the length of the samples instead of adding new samples for each pedal position?

[pantonality]

I\'m not convinced that the damping the string method is entirely accurate either. If the string were to continue sounding the result would be different than if the string is damped in terms of sound energy causing resonance. If there were a way to subtract the nonresonance sound from the resonance sound you would get a useful sample that could be used as an additive effect to give the resonance sound. The difficulty with this is the phase lock issue, but you only have to do it once for each note and dynamic level sample.

Steve