Procedural audio for rolling dice

Question

I'm looking for suggestions on how to approach randomizing the audio around rolling a pair of D6 dice in my game. I'd like each roll of the dice to sound different but be reasonably plausible. And I don't want to make a bunch of pre-recorded dice rolls, I'd like more variation. Note that I do not need to synchronize the audio with animations (the animation I'm using is very simplistic and abstract).

My assumption is that I need a couple basic audio snippets for the sound of a single die hitting a surface once. (Any suggestions for generating those? Or is it best to capture them?) Then I would need some way to mutate and combine variations on that basic sound to create a unique roll sound.... Or am I just too ignorant to understand how complex that would be and I should try a different approach?

I'm developing in Java for Android, but tutorials or descriptions of how to reasonably combine or procedurally generate audio in any language would be appreciated. I don't need real-time support as I think I could just generate the next roll's audio in advance and cache it until the dice are actually rolled.

Answer 1

The dice could generate 2 types of sounds - one when hitting the environment; and one while hitting the other dice. Reading a spectrograph of a dice collision sound could give you the ratios of overtones. Its easy to prototype this in a software like Pure Data. The idea is that by varying the fundamental frequency slightly you should have a more procedural collision sound. You can probably use random numbers to do the variation and predict collisions. Its probably not the perfect sound, but it could be a start.

This is probably very vague, but I hope it still helps :D.

Answer 2

My solution would be to use free recording software to capture the sounds of one die rolling at a time. The right software would be able to chop those files into small samples containing each percussive hit and the residual audio. Just a few recorded dice rolls would give you dozens of samples of "hits".

Then, using Java, load up references to each sample, and design an algorithm that would play them back in a semi-random sequence with appropriate timing modifications. Then you'd have plenty of variety on the fly without the need to mix audio into a single stream before playback.

Answer 3

You can look over the procedural code for the "Shaker" class from the synthesis toolkit (STK). STK is a C++ library, but the procedure for actually creating the audio samples isn't too hard to pull out. There are lots of type of shakers offered by default. I expect you could poke in at the parametrization and make modifications if you wanted.