Linux Audio Driver Improvements On The Horizon

[ninez]

Linux doesn't need an even better audio subsystem, it needs professional audio applications that don't suck. The problem is that the flagship DAW is Ardour, whose v3 release has been delayed for about 3 years now. Ardour3 was going to finally bring MIDI support, which most people consider critical, as Cubase and every other Windows and Mac DAW since about 1995 released have had MIDI support. The other problem with Ardour is that it crashes more often than a blind man driving a Ferrari. Paul apparently doesn't know how to fix it, so he announced that he'll be releasing Ardour3 soon regardless of the issues with it so that they can start piling more features on top of their crashy piece of shit DAW, while whining about how $3000 a month in donations just isn't enough.

I agree that linux needs more proaudio applications and that DAWs in linux have been slow to get midi... However, I disagree with your comment on ALSA (or FFADO for that matter) not needing improvements. They are alright, but do leave _a_lot_ to be desired, in certain areas - to claim otherwise is well, quite dumb :\

from my own experiences, there are certain critiques of yours that i find to be somewhat off of the mark (and laughable!) 1st. Ardour3 has crashed _maybe_ 10 times over the last 6 months on my machine. (being built from git, every few days) - i would say twice a month (just under) isn't too bad for beta software! (in fact, far less than using several other daws in both Windows/Mac, at times - including Ableton Live, Protools and Logic 9). 2nd. Apparently, you either A). weren't able to absorb what Paul's post @ ardour.org regarding A3's release was _actually_ about (is english not your 1st language / you have poor comprehension?) ... or B). you are intentionally being dishonest, misrepresenting what his position and what that post was all about.

That post wasn't about 'paul not being able to fix a bug' - 'so screw it, we'll just release it' - it was about their current development model becoming *stagnant*. He is finding that the idea of 'feature freezes' is slowing down development (being as you are limiting development to nothing more than bug fixing) and later in the comments he goes on to explain that in some cases the odd bug would be solved by changes that they had planned for after the feature freeze / official release... So now, by deciding to move forward, they can start implementing some of the 'deeper' changes and get development rolling again, while also continuing to work on bugs ~ which 1) makes development fun again / keeps interest amognst devs and 2) in some cases will make solving bugs easy 3) benefits users who may be waiting for features that are planned for 3.1 (such as video-timeline, for example).

all three of these outcomes, are good reasons to change the development model. but apparently, you didn't understand even *one little bit* of what that article was actually about.

anyway, I think the improvements/fixes coming to 3.8 should be good (although, i am not sure when i will actually be using those changes, as i will have to wait until linux-rt is using 3.8+).

cheerz

[rudolph_steinberg]

The real question then needs to be asked: Why after 10 years can no one seem to get a pro audio application that doesn't suck together? Maybe the app developers aren't the root problem?

When I attempt to run Ardour3, it usually crashes before I can do anything useful. It doesn't crash because of Jack or ALSA, it crashes because it has garbage code full of coding errors, they would do better to scrap what they have and start over again, like Cubase has done several times in it's history.

Windows and Mac have a strong ecosystem with multiple DAW vendors, and legions of plugin developers. Most plugins and instruments for Linux pretty much suck too, but the reason probably comes down to the lack of good DAWs to run them in. In Windows and VST, there are about 1000 plugin developers for every DAW developer, and everything usually just works. In Linux that ratio is perhaps just 4 to 1, because more people saw the need for to write a DAW to run plugins in, than writing plugins with no reliable DAW to run them in, although none of them seem to be able to write that magical stable DAW.

I think if one good project can go the distance and compete with Windows, then more top talent from Windows and Mac will consider crossing over to Linux. As far as proprietary DAWs, Reaper seems to have renewed work on a native Linux port, which could be it, but then again the Reaper devs have a legacy of not finishing what they've started, broken promises and so on, so I wouldn't wait on them. Nor would I wait on Bitwig Studio, I'm beginning to think they're not serious, they've been in beta for a long time, and seem long on dreams and short on actual product. I'm pinning my hopes on some of the newer projects that began as native Linux applications.

[GreatEmerald]

Except that OpenAL is not a driver? It's like OpenGL in the audio world, except it's proprietary.

[DebianLinuxero]

A lot still to be done

A lot of years passed since the first hardware accelerated sound card appeared (the SB Live!) and still there isn't a Linux implementation that takes advantage of that via OpenAL.

Nor Creative with Live!, Audigy, X-FI, nor Cirrus Logic with Crystal Media chips on Maxi Sound Fortissimo, nor Asus with Xonar ...

No company put effort on that, and no comunity effort on reverse-engineering.

Seems that reverse-engineering graphic drivers worth, and sound ones not.


Now, in graphics drivers we are not too far Windows, but sound ones need more coders.

[Adriano ML]

But we still require to use a beast like JACK to do "professional" audio under Linux...

And what is the problem here? I mean, jack is an extremely powerful tool and sound server. You could complain about it not integrating well with pulseaudio for now and thus not being user friendly, but it's nice to have a specialized sound server just for professional audio that takes no compromises.

And well, hardware mixer support is limited to a few cards (most of them Creative ones). And let's forget about advanced features of DSPs, like the E-MU ones or others.

And in what cases is this relevant for professional audio? You would never want to leave jack for your DSP chain because it's hundreds orders of magnitude more flexible than any hardware DSP and avoids costly latency and bandwidth usage between I/O devices, and hardware mixing is totally inviable while you are already doing all the DSP in software inside jack or whatever you use in windows/mac (guess what, there is no use for hardware DSP and mixing there too).
Also, nowadays it's cheaper to mix everything in the cpu than to use costly bandwidth to move dozens of audio streams to a very inflexible piece or hardware just to offload mixing. And even if you are talking about these hardware Equalizer, virtual surround or something you get with cheap HDA drivers in windows, they are pretty terrible in terms of quality and implementation. I would rather use a good pulseaudio plugin for system-wide equalizer and virtual surround, but alas, nobody is doing that right now and your best bet is to dump pulseaudio into jack and put your favorite EQ and other DSP in the path. This is actually a really good solution with very low latency even for games and high sound quality, BUT, it's not user friendly to fiddle with jack and make pulseaudio cooperate(as of right now).

MacOS X and even Windows still has an extremely high amount of advantages in the audio stuff.

Not "extreme", but I do agree their approach is more user friendly. Although not necessarily technically sound.

I really hope the KLANG concept gets matured soon, this is one of the lots of lacks for making Linux a proper desktop system.

As many much more capable developers than me or you have said, KLANG is not viable. It actually goes against from a lot of things made in mac and windows sound system, that you seen to favor.

[ChrisXY]

Internal speakers of clevo p170em doesn't seem to work anymore with this, just external speakers.
Realtek ALC892

Let's see if this needs to be a bugreport...

Well, one suspend later it works. Seems to be one of these usual bugs with this hardware.

One problem I only have with this notebook is that when I change pulseaudio's "master" volume randomly this happens: http://ompldr.org/vZ296MQ/gibberish.ogg and I have to change the pulseaudio audio profile from 5.1 to 4.1 or similar, then it works just fine again.
Maybe it's a hardware bug, maybe it's in pulseaudio, who knows...

[gamerk2]

Linux doesn't need an even better audio subsystem, it needs professional audio applications that don't suck. The problem is that the flagship DAW is Ardour, whose v3 release has been delayed for about 3 years now. Ardour3 was going to finally bring MIDI support, which most people consider critical, as Cubase and every other Windows and Mac DAW since about 1995 released have had MIDI support. The other problem with Ardour is that it crashes more often than a blind man driving a Ferrari. Paul apparently doesn't know how to fix it, so he announced that he'll be releasing Ardour3 soon regardless of the issues with it so that they can start piling more features on top of their crashy piece of shit DAW, while whining about how $3000 a month in donations just isn't enough.

For that matter, any of the other Linux DAWs with about 10 years of development that still aren't stable and don't have basic features aren't going to be the one to put Linux audio on the map either. There are a number of promising new projects out there, but only time will tell if any of them make it, or become just another abandoned Linux audio project that once had potential.

The real question then needs to be asked: Why after 10 years can no one seem to get a pro audio application that doesn't suck together? Maybe the app developers aren't the root problem?

[Bucic]

The only reasons people say that is due to historical reasons (early PA versions were not very good, but current ones are very good) and the fact that they are too lazy to configure it. And then there are a few oddball configurations whose drivers in ALSA are poor and PulseAudio just tends to hit bugs in them more often.

In short: if it works for you, definitely do not remove it. If it doesn't work, then configure it to make it work.

I do absolutely no fancy things re sound. Thanks! Very informative.

[schmidtbag]

The only reasons people say that is due to historical reasons (early PA versions were not very good, but current ones are very good) and the fact that they are too lazy to configure it. And then there are a few oddball configurations whose drivers in ALSA are poor and PulseAudio just tends to hit bugs in them more often.

In short: if it works for you, definitely do not remove it. If it doesn't work, then configure it to make it work.

I agree, I used to not be fond of PA. It always worked for me, but it was a little too heavy for my liking. But today, it seems fine. I don't do enough special things with audio to use it over plain ALSA but PA does greatly help when it comes to muxing, because ALSA is a pain in the ass when it comes to that.

All i want is a decent user friendly way of muxing channels in ALSA. The asoundrc files don't work half the time and their documentation is overall poorly written.

[GreatEmerald]

I'm switching from Ubuntu to Lubuntu. Would you advise a simple user to avoid PulseAudio? Why? Lubuntu comes without it but I need recorder and equalizer and the ones I know require PA...

The only reasons people say that is due to historical reasons (early PA versions were not very good, but current ones are very good) and the fact that they are too lazy to configure it. And then there are a few oddball configurations whose drivers in ALSA are poor and PulseAudio just tends to hit bugs in them more often.

In short: if it works for you, definitely do not remove it. If it doesn't work, then configure it to make it work.