"PulseAudio Is Still Awesome"

[Remdul]

If Linux wants any chance of being a good, reliable desktop, it needs to trash ALSA, PulseAudio, JACK, and maybe even OSS and write a kernel and userspace sound system that is both simple and coherent, yet flexible. I shake my head at the Linux kernel and distros of GNU/Linux for making so many beginner mistakes.

This. I'm afraid any other attempt at adding another layer, or re-implementing one, will only make things worse. A Wayland-ish approach would be nice. Just cut right through the years of accumulated cruft, cast it aside and get right to the modern and functional. And existing software that can't adapt to it, is thusly not being maintained properly, and therefor can't be that important anyway (if it were, someone would step up).

Until then, would also wish ALSA/jack (or whatever is to blame) would stop spamming stderr whenever I run my programs, grrr...

[89c51]

Yep. That's because that's where software dsp (including mixing) belongs. It serves 90%+ of the users' needs. If you have the need for low latency, go with Jack. ALSA may have flaws, yes, but I don't think exclusively locking your sound card is one of them.

It shouldn't be that way. And thats why most people doing audio are on Macs. Because they can plug a POS mic to a Hi end audio interface and get things done without having to use 2 different servers.

[wagaf]

ALSA and PA both are poorly designed:

(...)

If Linux wants any chance of being a good, reliable desktop, it needs to trash ALSA, PulseAudio, JACK, and maybe even OSS and write a kernel and userspace sound system that is both simple and coherent, yet flexible. I shake my head at the Linux kernel and distros of GNU/Linux for making so many beginner mistakes.

With PulseAudio the sound system in Linux is good enough for end-users and most professionals. It is actually simple, coherent and flexible.

For sure ALSA is shit, but if a new sound API was to be included in the kernel, PulseAudio should definitely stay there in userspace and use the new API simply because it's the most advanced and complete Free sound server available. The only value provided by ALSA is the (poor) kernel hardware abstraction layer but PA does everything else and does it right.

I've found PulseAudio to be an excellent piece of software at all levels. It works much better than every other Linux audio library for almost every possible usecase. As a developer I also enjoyed the excellent sound API they provide. For instance with a single line I could enable audio cancelling and let PA determine if, with the current setup, audio cancelling is necessary, and potentially enable it live (while my app is playing) if the user plugs something etc. No lower-level system have access to the necessary information to do this : it should really be done in userspace unless the whole desktop is put in the kernel. Note that OS X and Windows now also include similar sound servers.

[Dasher42]

Once upon a time, I ran Gentoo, because I was fed up with ALSA software mixing, ESD, and all that. It took USE flags to make sure no software mixing was loaded, and I then made sure I got mixing in hardware. In the days of a single-core processor, that means nearly double the framerates in games, and a much smoother experience in multimedia.

Ten years later, I have had quite a time getting JACK and PulseAudio to play well together, and I don't want to have to use Ableton on a Mac. I want to support open source and use it exclusively. However, the currently available Linux ecosystem is doing an iffy job of supporting people who take audio seriously. It shouldn't be that way.

[wagaf]

PulseAudio Is Still Useless

"PulseAudio does not currently allow TrueHD or DTS-MA passthrough, this is a PulseAudio limitation and not a limitation of the Kodi implementation."
http://kodi.wiki/view/PulseAudio#Pul..._Configuration

DTS-MA and TrueHD are just proprietary lossless codecs (like FLAC, which is free). They don't improve/add anything over PCM as far as audio quality is concerned (PCM is just plain uncompressed audio).

This means you will get the exact same audio if the DTS-MA track is decoded by Kodi then sent as PCM to your amp.

If for some reason you really need those feature feel free to write a commit.

But calling PulseAudio "useless" because of this is not reasonable.

[mdias]

It shouldn't be that way. And thats why most people doing audio are on Macs. Because they can plug a POS mic to a Hi end audio interface and get things done without having to use 2 different servers.

The fact that the Mac hides it's implementation doesn't mean it doesn't have 2 or more layers before reaching the hardware. I'm not familiar with the Mac stack, so I can't comment further in this respect.
What you're asking for, however, sounds more like a feature request for pulseaudio. I will agree with you here, it would be nice to have a "low latency" mode or better integration with Jack. But then some users would just complain more since it would add to the complexity of the server...

[edit] Take the graphics stack for example; we have the drm layer before libgl. And we have the compositor before whatever toolkit your apps are using. It wouldn't make sense to have your apps talk to the drm layer directly or draw directly to the screen now, would it?

[Marc Driftmeyer]

I felt embarrassed just reading that blog oh god... But I'm glad things work for him.
But things work even better if you remove unnecessary level of complexity and simply use ALSA :P

Alsa is a POS. Always has been. It's just the entire Sound ecoystem for Linux has been a POS hodge podge and people don't like a PulseAudio for fixing and simplifying it across desktop environments inside Linux. I haven't touched PulseAudio configs for nearly 18 months. It just works.

[Marc Driftmeyer]

On Debian Sid ( I won't switch to Fedora any time soon 10q ) it's the default yet I can't have sound in more than one app at once, believe me a tried to follow all the guides, to no avail.
Also none of the pavucontrol / manager apps actually registered any sound or app being used, no mpd, no Opera ( flash / html5 ), no mplayer, no sdl2 game, nothing.

I'm on Debian Sid and it works quite well. Don't have a single issue with anything you're mentioning. Been daily with Debian since 2001.

[profoundWHALE]

If Linux wants any chance of being a good, reliable desktop, it needs to trash ALSA, PulseAudio, JACK, and maybe even OSS and write a kernel and userspace sound system that is both simple and coherent, yet flexible. I shake my head at the Linux kernel and distros of GNU/Linux for making so many beginner mistakes.

Have you seen these?
http://www.phoronix.com/scan.php?pag...tem&px=MTE1MDc

http://klang.eudyptula.org/

This seems to be the goal of Klang, but no real news to be heard of so far.

[Luke]

"Most people doing audio are on Macs?" Not necessarity! I got my start on audio reporting and never had nor wanted a Mac. I always considered Macs to be hard to use clunkers, sometimes with no mic jacks, always with an unfamiliar interface. I had been used to first old versions of Windows on public computers, then to Linux machines with GNOME 2, in both cases using Audacity to edit sound. It was the natural free and open source replacement for Cooledit, with no paid version needed to output to mp3. Outputting from Cooledit to .wav back in 2004 made files too big for websites, outputting to .ogg made files nobody would open from a default Windows machine because their players didn't handle it by default. That was on a Pentium laptop with a 2GB HDD and Windows 95 on it. It was replaced by a far superior Athon 500MHZ/10GB HDD machines with what was probably Debian Woody on it later that year. Sure as hell XMMS was more reliable than Winamp for playback when running as an unattended sound server!