I have a friend who is an experimental musician. He doesn't program at all, but he wants me to write him a program. He is looking for a program which will produce audio output based on a more accurate mathematical model that standard MIDI. I already have the mathematical model, but I've never done any audio programming.

My friend is not acquainted with Linux, or any other *NIX except maybe OSX. Thus, I'll probably need the program to run under windows. However, I would like to write a portable program so that those of us with free software OS's can benefit. However, I can't find much in the way of portable audio programming information.

Is there any way to generate really raw audio output and play it back on multiple operating systems? I mean really raw - I'll be mathematically modeling the waveform and sampling it myself. I see that Java has a sound API, but I'm hesitant to use Java. He is likely going to wind up using this to model as many as a dozen strings simultaneously, and performance may become an issue as he'll require high bitrates.

Can I program it for ALSA, then run it through cygwin? Does it have an ALSA implementation? Any ideas?

Perhaps selecting a sound sever that plays on all OS's? That or just output raw PCM to a file, then based on the OS decide what to do with it.

I would just output the raw PCM to a file, but he'll probably want to play back in real time. Can anyone point me to a good tutorial on audio programming in general? I really don't know anything about the issue, and I might answer some of my own questions if I can find one.

To anyone looking at this - I found it. PortAudio. That's what Audacity uses to play back audio on multiple OS's. It just wraps the local audio interface.