Change Ubuntu default sound API from ALSA to OSSv4

Registered by LTSmash on 2008-12-04

Linux audio is a mess. We have a million APIs and a billion more sound servers. This is just totally unreasonable. Let's look at Pulseaudio and ALSA as that's our main discussion point:

ALSA: Undocumented mess of an sound system and 100% incompatible with other unices. Who wants to write professional audio applications specifically for Linux? Why target one platform when you can get other Unix systems working at the same time when you can target OSS and have it working everywhere? I've even talked to developers who run into broken/lying sound drivers for ALSA on a regular basis. It's not just commercial applications -- people these days generally care about having their audio work on both BSD and Linux. Some extend it as far to work on Solaris and others.

You have to remember guys, Linux isn't alone in this world. We rely a lot on developments of other Unix operating systems. By choosing to go the route of ALSA we're breaking compatibility and creating more issues. Reinventing the wheel isn't exactly something we should be doing at this point. Taking the current wheel and perfecting it is what 4Front's OSS implementation does, and is what we need to be focusing on.

Look at Pulseaudio. It has an amazing resume, but fails to deliver. Ever tried Hardy with Quake3(including ioquake3)? Quake4? ETQW? Those are specific applications I've found to completely and utterly fail under Pulseaudio. The sound is crackly. ETQW the sound is delayed by like 2 seconds. What happened to this "low latency" sound server? How are we to get low latency by adding MORE abstraction layers? It simply doesn't make sense.

I talked to Timothy Bessett, Linux programmer of ID software regarding ETQW and gave him some debugging output -- Pulseaudio is lying about what frequency it's using when it pretends to be a direct ALSA implementation. It's 100% on the Pulseaudio side. I tried contacting the main Pulseaudio developer and he gets all defensive saying his implementation is perfect, when clearly it is not.

There are constant reports of issues with Pulseaudio -- crackling audio, poor audio quality, latency issues, and things just not working anymore. Skype doesn't work very well, nor does Teamspeak. They might not be Open Source, but that's their decision. They're still applications people use every day. (Don't offer me alternatives, I know them all and also use them!)

Now let's try this with 4Front OSS.
Quake3, Quake4, ioquake3, Teamspeak, Skype (static/oss version), and more all work 100% perfect. No audio crackling, no latency issues. In fact, I can actually use Skype, Teamspeak, and Mumble at the same time for mic input! If I try to use Teamspeak and Skype the same time with ALSA, Skype crashes!

No audio quality issues? Check
Low latency? Check
All applications on my system working perfect? Check
Not breaking legacy support & compatible with other Unices? Check
Per application audio controls? Check
Open source? Check

It is completely irresponsible as a community to continue to use ALSA as our sound system. For those people who need it, Pulseaudio can be installed. Those people are going to be few and far between, because its only real feature now is network audio.

Please, Ubuntu devs, take this seriously. If you choose to do the next release with Opensound OSS you're doing the open source community a HUGE favor. Opensound OSS is growing every day in users and developers. They offer sound drivers (Xfi) that ALSA doesn't. Okay, so they have a driver (only one that I know of) they couldn't open source -- big deal. We'll work around it. It's what we always do.

Do the right thing and let's fix this stupid Linux audio catastrophe.


Blueprint information

Needs approval
Series goal:
Proposed for warty
Needs Infrastructure
Milestone target:
Started by
LTSmash on 2008-12-04

Related branches



Make OSSv4 compartible sith more hardware, add some of the missing features and change it to be the default sound system.


Work Items