Estimating Pitch & Volume of Collision Sounds

[Borek]

I need to develop some kind of simplified formula that will allow me to estimate pitch and volume of a sound of two colliding objects in a simulation. It doesn't have to be exact, mostly it is enough that it follows the intuition - large object produce lower sounds, large and fast object produce louder sounds. To simplify things let's say objects are spherical cows ('moo' doesn't count). Physical engine returns impulse of the collision, I also know mass and volume of the object.

So far I had some success with the pitch - I assumed that the oscillations of a solid object will be in a way similar to the oscillations of air in the Helmholtz resonator (one of the modes of the oscillations of a solid can be contraction/swelling, doesn't it?), that suggested frequency being proportional to V^{\frac {-1} 2} and after some tinkering with constants I am quite happy with the result - when I look at the collision and listen to the sounds it feels natural.

It is loudness that is a problem. I did some random checks making loudness linearly dependent on the impulse or impulse/mass ratio, but I don't like the effect. If nothing else works I will try to test exponential and logarithmic dependencies by brute force, but it will be time consuming. Do you have any idea about what the real thing is? Or some analogies that could be adapted?

 

[A.T.]

large object produce lower sounds

The stiffness of the materials involved also affects the pitch.

 

[Vanadium 50]

large object produce lower sounds

Is that true? It stands to reason, but I have heard some very low, almost thundery sounds from objects less than a half meter across, and much higher sounds from cars.

 

[Borek]

The stiffness of the materials involved also affects the pitch.

Definitely. I am assuming all objects are made from more or less the same material. Also for the same volume/mass something long and thin will produce different pitch than a sphere, but I am trying to keep things simple (hence the reference to the spherical cow).

Is that true? It stands to reason, but I have heard some very low, almost thundery sounds from objects less than a half meter across, and much higher sounds from cars.

Simulation combines initially stationary objects that start to fall or roll when pushed, and they are close to each other to allow interactions like in a domino effect, so the speed of the colliding objects is rather low. Hopefully that allows the simplification. But I am open to suggestions.

 

[256bits]

Your physics engine should be using something similar to this for ideal elastic collisions.
http://dbkcues.ru/articles-2/collision-of-billiard-balls/?lang=enSo knowing Young's modulus and Poison's ratio for the materials of the object, the time of interaction can be determined, and hence the impulse.

Yet no elastic collision is ideal, as evident by the energy given off by sound from the collision.
What mathematical relation impulse and the energy emitted by the collision have to one another as the sound will have an energy content, as well as the two bodies absorbing some energy as heat from the interaction.

Nonetheless, if that relationship is linear, or exponential, or whatever, the energy emitted with the sound may not be perceived again in a linear fashion with our ears. I am sure you have heard of sound decibel levels regarding intensity and pressure. Twice as much energy in the sound wave is most likely not seen as being twice the loudness, and is frequency dependent also if I recall.

I forget my sound knowledge so I can't give a better statement than that.
I t might lead you though with some keywords for a search on our ears perceiving sound.

 

[Klystron]

The author of this NASA publication develops three dimensional (3-D) audio simulations for human factor and crew studies. This sim includes spatialized simulated sound including speech synthesis and ambient audio but too long ago for me to remember coding details.

Probably more relevant if your cows talked to each other but may spark some ideas.

 

[OldManBOMBIN]

I'm not a physicist, but I am trying to pry into this particular corner for some ground-level research on a related topic.

I was wondering if you had taken into account air pressure around, or any sort of material properties of the bodies?

 

[Vanadium 50]

I very much doubt you can predict a sound from a realistic system based on first principles.

 

[OldManBOMBIN]

I very much doubt you can predict a sound from a realistic system based on first principles.

:-/
Well not with a mindset like that. But I can understand why you wouldn't want to pursue such a thing.

 

[berkeman]

Welcome to PF.

:-/
Well not with a mindset like that. But I can understand why you wouldn't want to pursue such a thing.

Please check the attitude at the door when entering. Thank you.

I was wondering if you had taken into account air pressure around, or any sort of material properties of the bodies?

Air pressure would have very little effect on the sounds, at least for pressures around normal atmospheric pressure. The properties of the materials and the construction would have strong effects on the sounds of the collision(s), but how are you proposing to do this simulation? With some sort of FEA?

 

[OldManBOMBIN]

Welcome to PF.Please check the attitude at the door when entering. Thank you.Air pressure would have very little effect on the sounds, at least for pressures around normal atmospheric pressure. The properties of the materials and the construction would have strong effects on the sounds of the collision(s), but how are you proposing to do this simulation? With some sort of FEA?

Sorry, I meant no aggression or anything. I was being a bit sarcastic at the first, but I genuinely meant the second half.

Honestly I'm just trying to flesh out my understanding of how things work, and how one might go about simulating these things - kinda see what has already been done, and ask questions on 4 year old forum threads.

My background is in automation and engineering, so I'm pretty much just googling stuff and shooting in the dark.

 

[Borek]

I was wondering if you had taken into account air pressure around, or any sort of material properties of the bodies?

I was looking for something that would give just results sounding natural, not for an exact solution.

So no, I didn't.

 

[OldManBOMBIN]

I was looking for something that would give just results sounding natural, not for an exact solution.

So no, I didn't.

10-4

 

[berkeman]

I realized overnight that this might be a good application for Neural Network processing and machine learning. You could feed a bunch of data in (information about the objects crashing and the sound generated), and see how long it took for you to feed something new in and generate a sound.

It's a very complicated problem in the general case, but you might be able to start with simple shapes and materials and see how it goes.

 

[OldManBOMBIN]

I realized overnight that this might be a good application for Neural Network processing and machine learning. You could feed a bunch of data in (information about the objects crashing and the sound generated), and see how long it took for you to feed something new in and generate a sound.

It's a very complicated problem in the general case, but you might be able to start with simple shapes and materials and see how it goes.

Excellent thought