Sympathetic resonance

DSP, Plugin and Host development discussion.
RELATED
PRODUCTS

Post

hugoderwolf; I agree that it is a very interesting topic, although the OP seemed to be looking more for advice on how to achieve a passable result rather than a deep discussion on the inner workings.

I believe anyone serious about keyboard instruments who has played a variety of pianos (just visit a good size >40 shop!) should understand the huge differences in instruments. There are so many complex factors at play I suspect even the most skilled engineers designing such instruments regularly discover something new.

Regarding your comment about coupled resonators; this is exactly what I was referring to when I described a simplified model based upon "wave-guides" (delay-lines) is only a very poor approximation. Not only are resonators at play, every component of the instrument including the body and components of the action right down to the keys themselves play an influence. You mention absorption: some components might be thought of as creating spectral notches rather than peaks, essentially doing the opposite people would at first assume to be the case with a resonator like a string.

There are of course always questions of diminishing returns. Then again for the people interested in this topic it hardly matters whether some random sample will distinguish A from B in a double-blind test or so on. Being able to examine the amazing emergent complexity resulting from such a seemingly simple instrument can be extremely fulfilling in itself.

I believe a conscious fine-tuning of many of the components such that they "line up together" in a well made piano is what contributes to the differences in sound via reinforcement of these effects.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.

Post

If you mean the string resonances produced when the sustain pedal is down, you can simulate it by recording an impulse response of the piano with the pedal down.
Apply the impulse on pedal down with a fade-in and cut it when the pedal is up.

If you mean the string resonances produced when you have individual notes held down and you play the same note or the 5th an octave higher or lower... that would create copies of the held note at lower volumes.

Post

flakes2 wrote:Physical modeling would be possible, but this is overkill on a piano
Why overkill? Plugins like Pianoteq show that it makes very sense.

Post

Please continue arguing via PM. Lets get back to the topic:
I was asking for an efficient way to simulate sympathetic resonance.

Post

Define efficient... And maybe you should ask Moddart, although I doubt they will unveil their secrets to you :hihi:
Fernando (FMR)

Post

I've never tried it, but if you were looking for something that had the flavour of sympathetic resonance but without going the full modelling route, how about adding a little 'kick' to the envelopes of any currently-held notes whenever a new note is played, with the size of the kick depending on both the velocity of the new note and how close it is harmonically to each held note. That might get you something that sounds better than nothing.

For modelling the sustain pedal, I think a convolution reverb effect might be useful to simulate all the strings resonating.

Post

kryptonaut wrote:I've never tried it, but if you were looking for something that had the flavour of sympathetic resonance but without going the full modelling route, how about adding a little 'kick' to the envelopes of any currently-held notes whenever a new note is played, with the size of the kick depending on both the velocity of the new note and how close it is harmonically to each held note. That might get you something that sounds better than nothing.
Sorry, but I fail to see how this will come with something resembling sympathetic resonance :?:
kryptonaut wrote: For modelling the sustain pedal, I think a convolution reverb effect might be useful to simulate all the strings resonating.
This is the approach being followed by some sample library makers. It's simplistic, to say the least. The best I can say of it is that it's better than nothing. But to come up with something more realistic, you would have to sample an impulse for EACH note of the piano, and activate the proper impulses when each key is played and pedalled. And for that you'd have to find a way to balance the overall volume (just thinking loud).
Fernando (FMR)

Post

stratum wrote:A piano is something quite different, obviously. Yet from a customer perspective the fact that remains is that a piano recording does not sound like a piano through a pair of PC speakers. Through an headphone, it's much better, but not quite the same, something is still missing, a good hifi set at about the right volume level adjusted and positioned properly would be closer, yet in that case most people would prefer the sound of a real but somewhat cheap Chinese piano to a VST emulating the sound of, say, Steinway.
When modeling something, whether it's a piano, guitar, trumpet, drum kit or whatever the reference has to be a digital recording, not some live performance. Your point seems to be that it's better to listen to say a classic orchestra in a real hall than to listen to a CD recording at home. That may be true of course, but is ihmo irrelevant for both music producers & plugin developers.

Richard

p.s. I cannot reply to your PM because it seems you have disabled PMs.
Synapse Audio Software - www.synapse-audio.com

Post

Slightly pcking a very small nit from the first page, not necessarily related to sympathetic resonance--

It is near impossible to make all three piano note treble strings or the two mid strings "perfectly in tune". With skill a piano tuning hammer can be tweaked "real close" but perfect unisons would be very difficult. Moreover, even if you get a "perfect unison" then after all the other notes have been tuned then your first perfect unison probably isn't perfect any more, and will gradually become even less perfect with the passage of hours, days and months. Even if you tune the beast every day it is never in perfect tune. :)

Another problem is consistency of materials. Wound strings are difficult to make perfectly consistent, and this is true even with the un-wound piano wire. The un-wound piano wire comes on big spools. If you for instance restring a piano, where all three strands of a unison are consecutive pieces of steel wire off the same roll, the three strings will not be musically identical. A certain percentage of the individual strings will have an enharmonic beat EVEN IF ONLY THE ONE STRING IS UNMUTED. It is rather difficult to tune a beatless three string unison if even individual solo strings can have beats.

There are various piano tuner tricks to try to charm a string to have fewer solo beats. At worst case if a string is real dissonant, you can try replacing the string in hope that the next piece of wire off the spool will be better behaved. OTOH it might be somewhat an interaction between the harp, soundboard and bridge at that location. In which case the new piece of steel wire might behave as nasty as the old piece.

Post

fmr wrote:
kryptonaut wrote:I've never tried it, but if you were looking for something that had the flavour of sympathetic resonance but without going the full modelling route, how about adding a little 'kick' to the envelopes of any currently-held notes whenever a new note is played, with the size of the kick depending on both the velocity of the new note and how close it is harmonically to each held note. That might get you something that sounds better than nothing.
Sorry, but I fail to see how this will come with something resembling sympathetic resonance :?:
Well, the OP was asking for shortcuts that didn't go as far as physical modelling. Sympathetic resonance includes the vibrations of other strings in response to sound coming from struck keys.

In a piano, strings whose keys are not held down are damped so will only resonate a little. Strings whose keys are held down are not damped, so will vibrate sympathetically if a 'harmonically near' note is sounded. I suggest that one way to simulate this might be to re-excite the envelopes of held-down keys when a new note is struck, with the excitation level depending on the proximity (both physically and harmonically) to the struck note.
It would be relatively inexpensive to do, and would hopefully be better than nothing.

Post

I don't think we're sensitive enough to such tiny changes in amplitude to pick up that effect unless it was exaggerated quite a bit beyond the amplitude change due to harmonic peaks in other strings.

We are sensitive enough to 500 Hz - 5000 Hz to pick up slight changes in resonances due to it being our vocal range.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.

Post

What you could do instead is take into account lower strings which would resonate at frequencies where harmonics of the low string come close (is it a cosine raised to nth power shaped function?) to a harmonic of the higher strings.

You could have a precomputed table for each note. If the note released matched any resonant sustaining notes, the release could be sustained at a level computed via combination of the values in the table for time and frequency. A simple low-pass or high-pass filter could be applied during the "post-release sustain" to mimic the frequency selective effect albeit with only very poor accuracy.

The filter could be engaged with an equal power crossfade shaped by the release portion for the envelope of the solo note (as if resonance hadn't been accounted for.)
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.

Post

just a thought, and I'm probably wrong, but don't the recent yamaha clavinovas have a setting for sympathetic resonance that uses some form of convolution switched depending on which keys are depressed ?
whiteLABEL - now set free : whiteLABEL ||

Post

There is no SR when mixing tracks in daws and no one implements this...
samplitude is the best daw for me. To have studio like sound before asking questions on any audio forums in the internet please read the book by alex unlocking fx creative potential

Post

solomute wrote:There is no SR when mixing tracks in daws and no one implements this...
nor should there be.
my other modular synth is a bugbrand

Post Reply

Return to “DSP and Plugin Development”