PipeWire 0.3.77 (2023-08-04)
This is a quick bugfix release that is API and ABI compatible with previous 0.3.x releases.
Highlights
- Fix a bug in ALSA source where the available number of samples was miscaluclated and resulted in xruns in some cases.
- A new L permission was added to make it possible to force a link between nodes even when the nodes can’t see each other.
- The VBAN module now supports midi send and receive as well.
- Many cleanups and small fixes.
it’s amazing how rapid the development of pipewire into a “usable state” so that distro quickly put it as drop in replacement for “matured” pulseaudio
Removed by mod
I am shocked that Lennart Poettering, PulseAudio dev, but also Avahi and SystemD dev, whose name might frequently be brought up in conversations about interoperability, reliability, small ego, diplomacy, etc etc, might not be as good at coding an audio stack as legendary C64 demo coder and PipeWire dev Wim Taymans. Shocked, I tell you. Well, not that shocked.
Let’s be fair here. PipeWire is leaps and bounds ahead of pulse and I was super happy when I could drop pulse. But pulse predates PipeWire by a decade and introduced concepts that were previously rather complex in Linux. It’s no coincidence its interface was adopted so quickly by audio tools and that it’s the recommended interface for PipeWire today (until devs are comfortable with recommending their own). Lennart saw the need and provided a solution which, in retrospective, could be much improved - but until PipeWire, nobody put in the work.
I too had my fair share of issues with PA. But it also solved some fundamental ones for me. I don’t miss meddling with .asoundrc or whatever it was to get dmix working. Pulse should not be measured against PipeWire, but rather ALSA, OSS and the stuff that the DEs brought with them (aRts…). It wasn’t always pretty. Early Pulse, however, wasn’t either.
Also, audio was originally not even in scope for PipeWire - it was touted as “PulseAudio for video”. So pulse didn’t exactly have a bad reputation even among PipeWire devs.
understandably, audio and graphics are different whole beast to manage, i guess
I didn’t even know Avahi is his project too, but was avoiding it too. Crazy how one person can create so many bad but popular projects.
I am curious, why is Avahi considered bad? Never used it so I am unfamiliar with the topic.
I have no idea, but on gentoo there is a use flag avahi and whenever I tried to enable it, it pulled so many dependencies I always blocked it.
I’ve read before that Pulse really had a difficult challenge, since it had to really resolve a lot of hardware vendor quirks that essentially would never be resolved. PipeWire gets the advantage of not having those early growing pains, because Pulse went through them. I’m not involved with the development of either to really know one way or another the truth behind that story.
Removed by mod
Why do you think these issues have nothing to do with drivers? Apart from confusing controls, all of these can be attributed to hardware and driver quirks.
Removed by mod
All of the following is speculation on my side and just stated as fact for easier reading / writing.
Because through the years I never had problems with plain Alsa on the same hardware
Because it did not allow for the functionality that exposed those bugs (as in PulseAudio provided features that should work according to documentation, but didn’t)
I don’t have problems with pipewire either on that hardware.
Because these bugs got fixed after PulseAudio exposed them.
Removed by mod
Sounds reasonable to me. The day only has so much hours. Software engineering is always a story of trade-offs, isn’t it?
Removed by mod
I love pipewire for my audio.
True, and it’s fabulous for video as well.
Ow you are right. I wasn’t aware of the video part of Pipewire. Which allows very low latency video recording as well.
Fun fact: with the proper plugins, you can capture both the video and the output of a particular window / process in OBS, giving you a stream of exactly that and nothing else (which you can then mix with other sources). Very handy for streaming if you don’t want to include voice chat (for privacy reasons) or background music (DMCA takedown on VODs).
However, for capturing Vulkan games, I recommend the vkcapture plugin. It acts as a Vulkan layer and is very performant, plus I get the impression it works better with Freesync.
I hate OBS under Linux still doesn’t allow me to use the GPU (using AMD videocard) for encoding.
Don’t blame OBS, AMD’s video encoding is awful in a multitude of ways
owh…? I was hoping it was open-source. Some standard API. or something…
As someone said, there are ways to get it working in OBS.
Does it expose the full feature set using VA-API? I have no clue. I was never able to actually hit the bitrate I wanted.
Please note that cards until the RX 6000 generation do not support B-frames.
AMD’s encoders are garbage, but if you insist, you can use VA-API to get accelerated encoding in the output advanced settings.
Perhaps a shot in the dark, but does anyone know if this update does anything to change the sound issues on Samsung laptops? For example, I have a GalaxyBook3 Pro 360, and it doesn’t matter what distro I use; I cannot get audio working. I spent 8 hours one day just going through forums and trying various fixes but could never get anything to work.
Is it because of missing firmware? From the Arch wiki on ALSA:
sof-firmware is required for some newer laptop models (mainly since 2019) because they implement their drivers with firmware provided by the Sound Open Firmware project.
I swear I tried all of that and uninstalled the existing firmware, and pulled the newer files. I might give it another go, but I saw on a thread that this is a specific issue with Galaxy Book devices. Someone had a detailed step-by-step guide to get the audio working; I tried it, but it didn’t.
Removed by mod
Yeah, it’s such a small percentage of users affected too. I have other laptops that run various distros of Linux without issues, but my Galaxy Book can’t get any audio unless I plug in speakers or headphones.
Thank you for the advice.
There’s a lot of love for it here, so I guess my experience isn’t typical. I updated to Ubuntu Lunar Lobster on my home media machine, which comes with PipeWire by default, and it’s utter shit. The vocals and some instruments in my music tracks only play nearly inaudibly from the center channel of my 5.1 surround system. It’s unlistenable, even with the center volume boosted.
Seriously, what am I missing? How can it do audio that poorly?
I’ve never had a multi-channel setup to experience configuring it, but I am sure there must be guidance out there. In my own experience PW improved my Linux audio experience quite a bit by resolving some issues I had with glitchy audio on my DAC and audio latency that was noticeable in some games.
