Submitted by “4Front Technologies is announcing the availability of Open Sound System version 4.0 for Linux, Solaris, FreeBSD, Open Server6 and UnixWare7. Open Sound System is a cross platform audio architecture that provides drivers for most consumer and professional audio devices and comes with an API that allows applications to be simply recompiled on any of the supported operating systems.”
In linux hasn’t OSS been replaced by ALSA?
IS there a reason why people would want to move back to OSS? or does this new release just help people who need to stick with OSS?
One of the benefits of commercial Open Sound System versus the in-kernel ALSA is that, like NVidia’s and ATi’s proprietary video card drivers, Open Sound System contains binary-only drivers for a lot of new sound hardware, such as Creative’s “new” X-Fi cards. Also, Open Sound System’s software-mixing supposedly works much better than dmix for those with older cards.
If OSS contains binary-only drivers, then it doesn’t sound so “open” to me. But, hey, why start this whole binary vs. open source argument again?
I really don’t get the point of having OSS be available for “Solaris, UnixWare7, and OpenServer6” as I doubt many people use these OSes on their desktops (and not many servers even have/use sound cards to begin with). If you’re a *nix geek that uses a *nix on his desktop, it’s probably Linux, so there’s little reason to move from the default (and much better supported) ALSA. An argument might be made that it would be nice to have a standard API for sound between the BSD flavors and Linux, but I don’t think we’ll ever see it (not to say that I’d be opposed to such an effort).
I remember back in my RedHat days (before Fedora came to be, and RedHat had a desktop distro called “RedHat Linux”), OSS was used, and things were bad…then, ALSA came; and things were good
Edited 2007-03-16 18:11
“An argument might be made that it would be nice to have a standard API for sound between the BSD flavors and Linux, but I don’t think we’ll ever see it (not to say that I’d be opposed to such an effort).”
Theoretically, an application can use GStreamer, then the Gstreamer could automagically use the right output plugin (see: autosink or gconfaudiosink, et al.).
Then again, this is much higher level and doesn’t work as nicely when you want to do lower-level sound hardware manipulation though…
“I remember back in my RedHat days (before Fedora came to be, and RedHat had a desktop distro called “RedHat Linux”), OSS was used, and things were bad…then, ALSA came; and things were good
“
May I ask which problems these were? I’m using FreeBSD for more than 5 years now at home at a regular basis, so it seems that I’m using OSS, and I didn’t have any problems related to sound (playback and recording).
>>If OSS contains binary-only drivers, then it doesn’t sound so “open” to me. But, hey, why start this whole binary vs. open source argument again?<<
Open as in “Open System” (think POSIX or Single Unix Standard) as oppose to “Open Source”. The point been that it provides a standard sound API that is “open” and works on multiple platforms. As an “Open System” one’s implentation can be under any license that you want as long as you follow the API standard.
Think of OpenVMS, they reason they changed the name from plain old VMS to OpenVMS was due to adding of Posix api subsystem (the “Open” bit)
If OSS contains binary-only drivers, then it doesn’t sound so “open” to me. But, hey, why start this whole binary vs. open source argument again?
There was a time when open was also commonly associated with *open* standards, and not just source code. In this case, OSS is open by standard, not by source.
I really don’t get the point of having OSS be available for “Solaris, UnixWare7, and OpenServer6” as I doubt many people use these OSes on their desktops (and not many servers even have/use sound cards to begin with). If you’re a *nix geek that uses a *nix on his desktop, it’s probably Linux, so there’s little reason to move from the default (and much better supported) ALSA. An argument might be made that it would be nice to have a standard API for sound between the BSD flavors and Linux, but I don’t think we’ll ever see it (not to say that I’d be opposed to such an effort).
Bzzzt.
There actually quite a few people who have workstations, etc. that use real UNIX on their desktop (not UNIX-like systems such as Linux, etc.).
Or, they use sound cards for which no public specifications are available and only OSS supports them.
OSS is for “Linux, Solaris, FreeBSD, Open Server6 and UnixWare7”, ALSA is for Linux only.
thanks Ford Prefect and codergeek42
that makes things a bit clearer.
ALSA does implement OSS emulation (and they aren’t removing it). So you can use OSS in linux too.
Or, more accurately, you can use OSS drivers via the usual ALSA interfaces.
In linux hasn’t OSS been replaced by ALSA?
Additional info from the article (so THAT’S what hyperlinks are for…):
Advanced Linux Sound Architecture (ALSA) Library emulation support so that popular ALSA apps (the ones that use the ALSA library interface) can run on Open Sound System.
IS there a reason why people would want to move back to OSS?
Yes, better sound quality in some cases. Better performance in some cases as well. Support, etc.
The fact that it also works on more than just Linux (unlike ALSA) is also a plus.
Some of us aren’t moving back to OSS, we never left
I personally will never use ALSA directly, it’s a Linux only sound API and is not portable.
The open-source Linux implementation of the Open Sound System APIs on Linux 1.x and 2.0-2.4 were horrible. Note: the Linux implementation was horrible, not the API itself.
All the other OSes that use the OSS APIs have had excellent sound (including in-kernel mixing of virtual channels) capabilities. Don’t knock the API because a single group of programmers couldn’t hack together a decent implementation.
ALSA is a Linux-only implementation. Why should the rest of the world be forced to move toward that, when there is already something better? There’s more than just Linux in the OS world.
Personally, I’ve never understood the Linux mentality of “let’s re-write it again” and “oh, we can (hopefully) do that better”, creating Linux-specific implementations, when there are already cross-platform APIs that do the same.
Improve the implementation, improve the design, get over the NIH syndrome.
ALSA/OSS is just API. It’s the driver quality that matters.
My VIA8237 chip on mb A7V880 produces high-frequency noise with ALSA, but not with oss’ driver. That’s enough reason for switching. Besides, the oss v4 also supports ALSA API via kernel-level emulation and low-latency sound mixing like jackd (may not be as good, but less problemic than dmix)
So, at least there is a way to get X-Fi to produce a sound on Linux. Although noone but gamers buy this card, and they aren’t really using Linux, it is still a bad situation and doesn’t help effort to grab a piece of desktop share. Maybe a new NDA scheme devised by Linux kernel hackers could be useful here.
And while their new soft mixing could be good, there is a good solution (which has also a good sw mixing) that I believe will become a main audio API: PulseAudio. (alsalib can use it as a backend btw.). Of course, it could be ported to work on new OSS, if it already doesn’t.
Xfi is released? Where?
http://opensource.creative.com/ says it isn’t.
No, it’s not currently supported with ALSA at all (as far as I am aware, the snd-emu10k1/emu10k2 don’t work with it). The commercial Open Sound System does contain proprietary (binary-only) drivers for the X-Fi though.
The commercial OSS drivers do not contain support for the XFi yet. This is something that is being worked on.
Odd. My friend, Alex, is using the OSS drivers for his X-Fi setup, or so he says.
Oh well; thanks for the clarification.
But if its software mixing capabilities really does have better quality and since it supports mixing of mmap audio, I might be really tempted to try it out.
But if its software mixing capabilities really does have better quality and since it supports mixing of mmap audio, I might be really tempted to try it out.
You can mix 5.1 surround with stereo and at the same time do full duplex voip and run mmap games.
Here’s what we tried in the labs with a ICH 4 device
– ogle playing DVD with dolby surround 5.1
– xmms playing stereo mp3
– ekigaa VOIP
– DoomIII (mmap)
Everything ran without any hiccups and audio was crystal clear. No asoundrc files to mess with, No other tricks – just run the apps as if you had complete access to the hardware.
Edited 2007-03-17 04:40
Sounds great, but you might want to improve the set up on Slackware systems, it looks like I’m still going to have to do some manual editing of config files.
Sounds great, but you might want to improve the set up on Slackware systems, it looks like I’m still going to have to do some manual editing of config files.
We have the tarball installation that should work fine as long as you have the kernel sources setup. THe only thing that may not work may be the automatic startup. But please file bug reports (www.opensound.com/support.cgi)
So now you can use ALSA on OSS on ALSA on OSS on ALSA…
Now what’s with those bin drivers ? I thought the name was “*Open* Sound System”
Was wondering about porting it to BeOS to bring in some more drivers, but it seems it’s pointless.
Open since years to different operating systems. Not open as in opensource or open like “free”, hey stop I do mean free not “free as in freedom” and so on. Do you get it?