So here's an idea on using computers for sound generation. Sound can be synthetically generated by setting a number of parameters to the right level. Alternatively, when a certain sound in real life can be reproduced, it can also be recorded digitally and edited as a file, a practice known as sampling. These are what I would regard the easy way - and admittedly versatile way - of generating sounds in the lab.

But here I would like to present a very different and ludicrously more expensive way to generate sound. It is possible to parametrically describe objects as well as model the sound they would produce upon contact with each other. Take for instance a rock and water. If both can be described in their appearance and behavior, such as stiffness and viscosity, as well as the sound it would generate when that particular rock, given its size, shape and weight, falls into that particular body of water, it will open up an entirely new approach to sound generation. I could imagine that a sound engineer when starting to collect sounds for a new track would instead visualize the objects needed that would yield a similar sound to the one he has stored in his data bank. He could then rather than browse for that sound, start to model the world in which those sounds are produced. And rather than tweak parameters to create a slightly different sound, the engineer tweaks the model of the mini-world: making the rock bigger or the water more shallow. Or add objects: two stones instead of one. Or change the medium, like increase its humidity or temperature.

It would demand a far larger computational effort than with current synthesizers, but perhaps is one that produces sounds more life-like than the sounds we hear these days.