KVR Audio

flakes2 · Post by **flakes2** » Mon Oct 24, 2016 8:10 pm

When a piano or a guitars plays the strings do also affect the other starings. How can sympathetic resonace be emulated?
Physical modeling would be possible, but this is overkill on a piano

stratum · Post by **stratum** » Mon Oct 24, 2016 8:32 pm

Physical modeling would be possible, but this is overkill on a piano

Are you sure those resonances would be rich enough in tonal quality to actually require physical modeling?

fmr · Post by **fmr** » Mon Oct 24, 2016 8:36 pm

stratum wrote:
Physical modeling would be possible, but this is overkill on a piano
Are you sure those resonances would be rich enough in tonal quality to actually require physical modeling?

On the piano, not only they are strong enough as also their lack alters the sound completely. But sample libraries usually sample sympathetic resonances as they sample everything else (when they are there, which isn't always the case).

stratum · Post by **stratum** » Mon Oct 24, 2016 8:46 pm

On the piano, not only they are strong enough as also their lack alters the sound completely. But sample libraries usually sample sympathetic resonances as they sample everything else (when they are there, which isn't always the case).

I do not doubt that they exist and are strong enough to be audible, but I suspect they may not actually be rich in total quality.

fmr · Post by **fmr** » Mon Oct 24, 2016 8:53 pm

stratum wrote:
On the piano, not only they are strong enough as also their lack alters the sound completely. But sample libraries usually sample sympathetic resonances as they sample everything else (when they are there, which isn't always the case).
I do not doubt that they exist and are strong enough to be audible, but I suspect they may not actually be rich in total quality.

What do you mean with "rich"?

stratum · Post by **stratum** » Mon Oct 24, 2016 9:06 pm

What do you mean with "rich"?

Some sound that is worthy to be modeled physically or be sampled because it cannot be generated using simpler methods.

For example, if you hit a string with a hammer, the initial sound is complex, but gradually overtones diminish and some simpler tone remains. I know at least one piano sampler software that actually claims to use this feature to combine sampled and synthesized sounds to reduce hard disk space, i.e. only the initial tone is complex enough to require sampling.
Now returning to the resonances, since they are generated by vibrations of nearby (close in frequency and/or distance) strings , chances are high there isn't strong energy transfer to initiate generation of a complex vibration, i.e. the complexity of the sound heard may be due to something other than the complexity of the vibrations of individual strings. That is, to me it seems like this requires a closer look before deciding that it is actually something that absolutely requires physical modeling.

EvilDragon · Post by **EvilDragon** » Mon Oct 24, 2016 9:10 pm

Pretty much half of the greatness of piano sound is exactly in sympathetic resonances. And yes, sampling is not getting there all the way, but physmod totally is.

fmr · Post by **fmr** » Mon Oct 24, 2016 9:21 pm

stratum wrote:
What do you mean with "rich"?
Some sound that is worthy to be modeled physically or be sampled because it cannot be generated using simpler methods.

For example, if you hit a string with a hammer, the initial sound is complex, but gradually overtones diminish and some simpler tone remains. I know at least one piano sampler software that actually claims to use this feature to combine sampled and synthesized sounds to reduce hard disk space, i.e. only the initial tone is complex enough to require sampling.
Now returning to the resonances, since they are generated by vibrations of nearby (close in frequency and/or distance) strings , chances are high there isn't strong energy transfer to initiate generation of a complex vibration, i.e. the complexity of the sound heard may be due to something other than the complexity of the vibrations of individual strings. That is, to me it seems like this requires a closer look before deciding that it is actually something that absolutely requires physical modeling.

Do you know what "sympathetic resonance" means? Any string vibrates sympathetically, if it is un-damped, because it creates "nodes" harmonically related with the impulse of the striked string or strings. Some vibrate more, some less, dpending oh their closer or more distant harmnic relation with the strings striked. But the harmonic board of the piano amplify these frequencies, which means that the strings are not simply being heard isolated in the air - their resonance is being amplified. And there are a lot of strings. The bigger and better the piano is, the more resonances are present.

These resonances vary according to the keys that are being striked, which means that their resonance is not uniform os constant. It varies constantly, according to the strings sounding. I don't know whatr you arwe trying to achieve, but if you are trying to emulate a piano in any way without string resonance, forget it.

This is difficult to describe. The better way is for you to listen the phenomen yourself in a real acoustic grand, or baby grand.

foosnark · Post by **foosnark** » Tue Oct 25, 2016 1:16 pm

One of my favorite ways to play with sympathetic resonance is with my HAPI Origin. Strike one of the tongues and the other 7 all resonate enough to feel; you can mute them individually with your hands or let them rattle on a hard mallet, etc. Honestly I think I love the feel of playing it more than the sound.

stratum · Post by **stratum** » Tue Oct 25, 2016 1:18 pm

Sounds like something I have expected it to be and yes, piano as a whole is a complex instrument.
That's exactly what I mean, trying to physically model it on a single string will not work (it's unnecessarily complex on that, as an approach, that is what is usually meant by a 'physical model') and unrealistically simple as a whole, when the whole instrument is concerned. So yes, I stand by what I have previously said. There are many strings on a piano, some notes have as much as 3, perhaps 4 (I do not recall the exact number), and each one is slightly out of tune and each one is not likely to vibrate in a complex way due to resonances from nearby strings. The whole thing is complex, not the strings. The initial hammer action creates complex overtones on the strings, that's what the best physical model -as found in the literature- will give for you. You could just sample that as well. As for the rest, it needs a closer look, maybe it's a lot simpler than what you think, if those simple 'parts' were correctly combined...

fmr · Post by **fmr** » Tue Oct 25, 2016 1:39 pm

stratum wrote:Sounds like something I have expected it to be and yes, piano as a whole is a complex instrument.
That's exactly what I mean, trying to physically model it on a single string will not work (it's unnecessarily complex on that, as an approach, that is what is usually meant by a 'physical model') and unrealistically simple as a whole, when the whole instrument is concerned. So yes, I stand by what I have previously said. There are many strings on a piano, some notes have as much as 3, perhaps 4 (I do not recall the exact number), and each one is slightly out of tune and each one is not likely to vibrate in a complex way due to resonances from nearby strings. The whole thing is complex, not the strings. The initial hammer action creates complex overtones on the strings, that's what the best physical model -as found in the literature- will give for you. You could just sample that as well. As for the rest, it needs a closer look, maybe it's a lot simpler than what you think, if those simple 'parts' were correctly combined...

"each one is slightly out of tune" - Only if the piano is not maintained. Each of the two or three strings have to sound perfectly in tune, otherwise you would have that "honky-tonk" effect.

"and each one is not likely to vibrate in a complex way due to resonances from nearby strings" - Take off the NOT. EACH string (meaning basically ALL OF THEM) is supposed to vibrate in a complex way due to the vibration of the STRIKED strings (not the nearby strings). Don't forget that all the strings are attached to the piano harp, and that one is physically attached and in contact with the harmonic board. All of this resonates.

stratum · Post by **stratum** » Tue Oct 25, 2016 1:48 pm

OK, whatever. Nobody will know the answer unless somebody finds a way to actually measure individual parts of the real thing.
Do not just solve the problem in the head, make some measurements and see what's actually happening, if possible. That's what I am trying to say. That can save a lot of time. Something to keep in mind before getting lost in that "physical" modelling literature.

Miles1981 · Post by **Miles1981** » Tue Oct 25, 2016 1:52 pm

Sympathetic resonance is known to be the missing elements in lots of piano models and it does make a noticeable difference.
The fact that the usual string model is made in 1D, and not 3D, not taking into account the air/wood displacement and the fact that this displacement will make an effect on the strings could explain some of the effect missing.

EvilDragon · Post by **EvilDragon** » Tue Oct 25, 2016 2:10 pm

fmr wrote:"each one is slightly out of tune" - Only if the piano is not maintained. Each of the two or three strings have to sound perfectly in tune, otherwise you would have that "honky-tonk" effect.

Actually, not necessarily

Each of the three strings can (and usually do) get VERY VERY SLIGHTLY detuned to make it sound better/thicker/phatter/what have you

Only if you go too far do you get the honky-tonk effect. Not to mention the very often used stretch tuning, which slightly detunes octaves as things spread away from middle C, but that's a different matter altogether.

Miles1981 wrote:Sympathetic resonance is known to be the missing elements in lots of piano models and it does make a noticeable difference.
The fact that the usual string model is made in 1D, and not 3D, not taking into account the air/wood displacement and the fact that this displacement will make an effect on the strings could explain some of the effect missing.

Well, Pianoteq models them in 3D, and it does account for air as well (you have "speed of sound" parameter even).

whyterabbyt · Post by **whyterabbyt** » Tue Oct 25, 2016 2:14 pm

stratum wrote:OK, whatever. Nobody will know the answer unless somebody finds a way to actually measure individual parts of the real thing.
Do not just solve the problem in the head, make some measurements and see what's actually happening, if possible. That's what I am trying to say. That can save a lot of time. Something to keep in mind before getting lost in that "physical" modelling literature.

You mean, do all the stuff you'd be doing for a physical model anyway. :sigh: