For this other little project I was working on
http://sciprogramming.com/community/index.php?topic=1618.msg9332#msg9332I thought about how to implement audio, and I was going to offer support for sciAudio (e.g. just make it work automatically, because why not? I could do so by intercepting file system calls and seeing if it's writing to the conductor file, etc....).
But looking at what it takes to set up sciAudio and how it works, it seems like it could be simpler. It looks like most of the functionality could be abstracted away in the Sound class itself, and you wouldn't need to special case "sci audio" sounds. Your wav/mp3 files (it looks like they need to be renamed to .sciaudio, but I'm not sure why that decision was made) could just be numbered (or prefixed with a number) and plopped in the main game folder. e.g. 104_GunShot.wav. Then when you play sound 104, it would check to see if there is a file corresponding to that number, and send the info over to sciaudio instead of calling the DoSound kernel. Or rather, calls to DoSound could be replaced with calls to DoSoundCustom, which either forwards to DoSound or sciAudio. Basically, it would mean supporting all the DoSound functionality and forwarding to sci audio instead.
Of course, there are some features like fadein/fadeout times that the Sound class doesn't support (though it could be extended to do so).
Anyway, I just thought of this as I was looking how to support sciAudio - and instead I realized it would be much easier to just support wav files in the DoSound kernel - it's a little more seamless to write scripts against (and it's basically what Sierra did when they started supporting digital audio). And then I thought why couldn't sciAudio do the same? But I haven't used it, so maybe I'm overlooking something? If Gumby pops in again, maybe he could comment? Or maybe someone wants to take this up on their own? It might only require changes to the sci scripts.