Linux sound is a steaming pile
renesd.blogspot.com
renesd.blogspot.com
X handles Mouse and Keyboard and Screen I/O, and there is no reason sound should not be considered a basic I/O that should go too.
Just add another
Section "InputDevice"
or xorg.conf, and OutputDevice as well. Each Xserver will list the specs it can handle (frequency/bitrate/channels), and clients will either adjust themself to the capabilities of the server, or pick some master format and convert on the fly as needed.http://opensoundcontrol.org/introduction-osc
JACK is pretty good too...but I see your point. Audio development on Linux fragmented several years ago - LADSPA vs ALSA was the start of that, IIRC - and although it's fairly robust today, almost everything feels years behind the Windows/Mac platforms. Video is even worse. For anyone but experimental artists or those with an in-house development team, it doesn't offer much of a platform for commercial work and is likely to scare potential clients away.
95% of the desktop users needs sound to work [a] when two streams (e.g. a video game and music) are played [b] low latency sound in video games (a.k.a lipsync)
Another 4% wants good audio tools to work on in Linux - which there arent many of. I dunno if this is a catch-22 because of the innumerable sound SDKs.
In a previous comment I had posted described Jonathan Blow's (creator of Braid) problems with low latency sound and APIs on Linux. And this is a guy who has been using linux for over a decade now.
The thread over at his blog has been going for over a year now (http://braid-game.com/news/?p=364) and makes for interesting reading.
I installed Rosegarden on Ubuntu 9.04 with high hopes, but after hours of monkeying around with it, it flat-out will not play music. Likewise, I installed timidity and what I had read was a kick-ass soundfont, only to discover that about a third of the instruments are simply missing.
Arrgh!