[bitL]

Just a question - if I think about flexibility, I think about composing harmonic functions directly in the code. I understand that historically synth music was based on analog components, and as people got used to these even virtual analog was born to mimic old analog stuff. Did anyone think about switching to a higher gear and just do pure math by composing functions directly without specials like "oscillator", "LFO", "envelope", "filter" etc.? Main issue/difference between synths is how they enable to compose these operators/components together and that's where we talk about how easy is it to program a synth, with implementation of individual components giving it sonic signature. If I look e.g. at Max4Live, this helps the compositing process, yet it doesn't help much with the use of completely arbitrary functions in your digital synth. So I am feeling we are missing out on a vast ocean of sounds never heard before.

[JoeDaDude]

The subtractive synthesis model which you describe, now done digitally, may be the most widely implemented model, but there are several other models that at least approximate what you are describing. For example Csound [1], which allows for lower level functions to be used in synthesis. However, there is a leap required to go from low level math to a functioning instrument that a composer or performer can use. [1] http://csound.github.io/

[surgesg]

plenty of people experimenting with what you're describing - see the work of Xenakis, Roads, Brun, etc.