Pipewire started life as a way to handle the gaps left in Wayland for stream capturing with v4l2 and dma-buff, zero copy screen sharing, audio/video sync, etc. They ended up recreating the pulseaudio server as a plugin and reimplementing the JACK interface with a focus on latency. It really is a wonderful project.
Great info! Thanks!!! Have been using Linux for over 15 years, never gave a second thought to how audio works on Linux. After watching this, I uninstalled PulseAudio and installed PipeWire. I have a large living room, and used to get "crackling" when using my Bluetooth headsets. This is no longer the case with PipeWire!! Have only had it installed for an hour though, so will see how it goes.
@@Lambda_Ovine I'm loving it! I'm not really an "audiophile", but it does sound better on my setup (no more intermittent crackling). I've converted my laptops over to Pipewire too :) Does everything I need to do, but my needs are very straight forward and simple (i.e. just need audio through Bluetooth speaker). Actually, I WFH, so I video conference too.
Actually, I noticed the crackle/stutter under PulseAudio on my laptop as wel last yearl, I had tried so many things with Pusleaudio such as forcing SBC (thinking it would use less bandwidth or something), messing with the server delay and buffer times (thinking it was a buffer underrun or something), even turning off WiFi in case both the 2.4 GHz signals are interfering... I guess this did disappear when I switched over and never noticed, neat!
An update to this video, taking into consideration the recent audio technology updates (using ALSA and PIPE WIRE only), would be very welcome. Tank you for this video anyway.
Thanks for doing such a nice rundown. I have been using PipeWire for a good few months now without any major issues. I had major issues about half a year ago (I know this video was released even before that time). In my experience so far, BT headphones work better with PipeWire than they did with PuldeAudio, though I don't use them that much. An amazing thing is that using PipeWire I can record audio from my BT headset right into Ardour, like any regular audio input. Plus I can work on my music using them in playback only profile. It's pretty amazing.
Yeah, I think the stability and support for Pipewire has greatly improved in the past year. As for recording audio from Bluetooth, make sure you check which sound profiles you're using, since some definitely sound better then others with Bluetooth (SBC-XQ vs SBC for playback for example)!
Really excellent explanation and exactly the kind of patient, thorough guide through linux audio that I never knew I wanted. I recently had some trouble switching my audio output from the hdmi port on my video card to a usb out connected to my headphone amplifier and suddenly I needed to learn about how sound worked. You seem really knowledgeable and did a great job explaining the full stack of technology powering audio. I would absolutely tune in for more videos like this (even if they take a year to put together :-)). Thank you!
`pavucontrol` is probably the easiest DE-agnostic way to easily switch inputs and outputs in general or on a per-program basis provided you're using PulseAudio or PipeWire.
I'm a total noob when it comes to linux however If I understand this correctly: Pipewire has the potential to give us pro audio users the ability to route multiple sources to the same and/or different outputs simultaneously. If this is the case, this would be game changing for those looking for advanced audio routing solutions. It would also be much more affordable than using Dante where you would require something like a CTP DIO88 Analogue to Dante Converter upwards of ($1000/£1000 give or take) to convert your devices and send them where you need to.
Stumbled upon this video while searching for a way to get pipewire to output 5.1 surround audio via S/PDIF. Very informative and while it didn't solve my problem, at least I now know more about PulseAudio and Pipewire. Researching resumes...
Hm, you are probably right. I remember that 5.1 wasn't fully implemented working in Pipewire last year. While I don't have an SPDIF output or a 5.1 setup to check on my current system, I checked the supported HDMI profiles and it does allow for a 5.1 or 7.1 output (as well as allowing 5.1 digital output on my old SoundBlaster), so it might be possible using HDMI instead if it isn't supported over optical.
I have a 5.1 sound interface connected via USB. All the 6 channels show up for separate connection in qpwgraph, and I was able to route audio to them separately.
This was a really great explainer. I never got how ALSA worked until now. I'm still unsure about Pipewire since it seems so buggy, but hopefully it will be stable enough to try on the next Mint LTS. After all those Ubuntu users test it for us, lol.
Ah yes the Arch/Fedora/SuSe Leap -> Ubuntu/Manjaro -> Debian/CentOS/Mint chain of software propagation... Jokes aside, thanks for the kind words and glad you found it helpful!
You are an amazing teacher! It is extremely rare to have capacity for knowledge and a great personality. If only grade school was full of interesting people, everything might be different. Thanks for sharing this knowledge strangely two months before I needed it.
Linux Audio is why I daily drive a hackintosh, despite being very keen to switch to Linux. I'm hoping that as Pipewire matures everything gets a lot simpler and easier. Thank you for the video, at least I now have a better grasp of the issues.
Glad you found the video useful! I remember I thought the same but for OS X audio when I had to use it for a job. Drove me crazy for a bit back then trying to find a way to capture both the system output and microphone input, which I could do on Windows and Linux.
Thank you for keeping this up over the years. It's people like you and content like this that made UA-cam the beautiful place it once was and make me reminisce. Thanks for making me feel nostalgic AND giving me the high level overview I was looking for. Keep on tinkering my friend.
This was very informative. Instead of you going like “yeah use this bc uhh yeah” you actually explained all the options and the pros and cons. Good stuff
great work, very informative! just a footnote: the minimum sampling frequency must always be at least 2x the max frequency being "recorded", otherwise spectral info could be lost, as stated by the Sampling Theorem (Nyquist-Shannon). so even if we can only hear up to ~20kHz we actually need at least ~40kHz in sampling frequency. in reality the distortion caused by the sampling process is almost imperceptible to most people but it exists, and human hearing ranges may vary as well
You are absolutely correct about needing at least double the frequency to capture the original information as per the Nyquist-Shannon sampling theorem - hence most sound cards running at 48 kHz these days. If you're interested, Chris Montgomery also has a great video exploring and demonstrating the sampling process as well as effects quantizing the input over on Xiph's website: xiph.org/video/vid2.shtml
This video helped me grasp things that I wouldn't have known how to search or troubleshoot. I've been able to solve a problem on my PC thanks to you! Appreciate the info 👍
I almote Broke my system while getting my Bluetooth Headset (Microphone) work. But I think I will give up on Bluetooth Headphones atleast for Meetings.
Finally I understand how alsa and pulseaudio interact, and how they relate to OSS! Thanks for the excellent video! Tips for future vids - more b-roll and perhaps slow down a bit
Thanks for the kind words! Early on, I used to try to get b-roll, but then I started to dread making videos (and I couldn't think of much to film for b-roll for a sound server), so I reverted to back simpler filming and editing... As for slowing down, yeah, it's something I need to work on, it kind of 'just happens' when I get a little passionate.
That was a first class system overview; your video really helped me to picture how it all 'bolts together'. Using qpwgraph, I've just set up the laptop microphone -> EasyEffects (so I can use eq, multi-band compression, etc) -> wfview (which then 'feeds' audio to and controls an Icom IC-705 transmitter on my network) and it's all working an absolute treat. All I have to do now is spend a ludicrous amount of time setting up all the EasyEffects components in an attempt to make it sound good(ish). 🙂 Assuming it sounds okay (long story about Icom codecs omitted to save space), the next stage will be to set all this up in a headless Linux box (with a decent audio interface and microphone) and try to emulate an Orban 9105A Optimod-HF using EasyEffects; I might be gone for quite some time! 😀
You are the man!!! You finally helped me understand how all these different processes work on my Linux machine. Your approach to explaining everything was genuinely balanced with simplicity --yet covering all the nuances of the interplay. Bravo!!
Thank you! Very well explained. I have a small music studio and Linux has become my OS of choice in the past year. My Linux skills have not yet been up to the task and I have had to use ALSA exclusively. Pipewire is intriguing me and I will give that a shot. I have one small suggestion for you. If you put a foam pop filter over that mic, your audio would be much improved. Have an awesome day!
Wow! I did not expect this to be so informative. By the end of the video, you have put every concept, you had introduced in the beginning, together so clearly. Thank you, really.
10:03 One reason PulseAudio makes things easier is it basically slots in under the ALSA API. So apps that know about ALSA automatically work with PulseAudio, without noticing the extra PulseAudio layer between them and the actual hardware.
Dude, you cleared a bunch of doubts I had in just 20 minutes! I bought an old Makie Onix-i 16 channel mixer that only works (so they say - I haven't tested) in linux. Also, the UbuntuStudio distro with the low latency driver is wonderful! A curiosity: in the old days of 100MHz CPUs, mixing audio was heavy for the processors and digital telephone switchers were built with those processors and requried not low-latency but real-time OS and programming. It was so difficult only 8 countries managed to design and produce them.
Worth every second of my life listening to this. Thumbs up for quickly orienting me on one piece of this confusing morass that is Linux audio technology.
ALSA can do software mixing. PulseAudio is not required for that. I listened to "podcasts"[1] while playing SNES emulators on Linux before there was PulseAudio. What PulseAudio can do is to give each program a different volume, resample, and route different programs to different output devices. And what PulseAudio actually did when it first arrived was to add such a huge audio latency that gaming and watching videos under Linux became impossible. Well, I had to uninstall PulseAudio to fix it. [1]: Wasn't called a podcast back then. Was mp3 files of Chaosradio manually downloaded from an FTP server.
I came here to say something like that. Dmix, dsnoop, plug, route and so on defined in asoundrc it rock solid if a bit cumbersome. There is also the possibility to setup ladspa plugins within the asoundrc settings file to e.g. split and filter audio for a satellite+subwoofer system.
Wow! You covered a lot of ground in one video, Tony--my head nearly exploded! Thanks for an informative march through the sordid history of Linux audio, most of which I have battled with at one time. Linux audio can be quite powerful, a Swiss Army Knife with 99 blades; just be careful when using a blade you don't cut yourself on one of the other 98 blades. Cheers!
I just Probed my machine Apple product: MacBookPro9,2Linux Mint 21.3 Virginia Ubuntu 22.04 jammy Desktop: Xfce 4.18.1 hoping to work in Katana 50 BOSS Tone Studio OMG! Wish me luck! Thank you for a very Streaght to the point instructional loved the pace of the info you covered. I have sndio , PulseAudio, ALSA, and PipeWire all running now JACK is not running. Working on it keep on keeping on! Thank You Very Much ~ Samuel 🙌💻
This is such a good video holy shit! I have been losing my mind over what these software were and briefly explained all their purpose holistically. Thanks!
A lot of ppl have over the years commented on the messiness and confusion around linux sound and could there be an easier way to deal with it. I have used alsa from the command line and pulseaudio via a gui. Not used Jack as I don't mix sound. Pipewire I've not heard of prior to watching this video. An instructive video which explains the basic differences between the various pieces of software and how they interact to provide sound capability for Linux which can be a bit overwhelming for Linux newbies. Keep up the good work.
I’ve noticed some sound quality issues on my Debian 12 install so I started researching how audio even works on Linux and which sound server to use. So the gist of everything is to use alsa + pipewire, gotcha.
Using PipeWire for almost two years now and I'm happy to announce that the BT device being "there" but not outputting any beep at all is back again :D Usually resolved by restarting the headphones 😩 From a gut feeling it's probably some chip going into energy saving mode after the last stream closed and it may be related to the headphones themselves. They did start to develop weird glitches over the time (especially the ANC feature) so there's that. Anyway, will totally stick with PipeWire. It really changed everything for me. And that's coming from someone who remembers how to echo "magic integers" into OSS device nodes to get sound going for some really odd native Linux PC games back in the days :D
Excellent video bro! I been doing Linux Audio off an on for a couple decades now and only learned enough to navigate and utilize these various apps you mentioned without ever really understanding their interactions. Just getting ready to give it a try again since winter is upon us already. That was super cool! Thanks!
Keep up the good work!!! There's a lot of information out there. Few people are able to dive deep into it and explain the way you did! Regarding a few comments about your hair... Embrace it!!! If i were you, i'd even hang an Einstein poster in the background. Awesome people try to help other's succeed. Hater's try to stop other's success! You're awesome!!! Keep it up Boss!!!
I love you, so cool how you know so much and are this passionate about the topic, it seemed so confusing it has been giving me headaches since I switched to linux
Hi Tony, thanks for all the info you've put together here. I've been editing and recording for over a decade on mainstream platforms, and only recently have started to delve into the Linux side of audio. All these bits and pieces of info are of great help! :D
Nice. Finally getting around to moving from xorg to wlroots and pulseaudio to pipewire... I'm a bit of a dinosaur haha so the time has come, been, gone and come again. This was really useful for clarifying some assumptions I'd made about... stuff. Cheers!
The Same Bluetooth problem happened with me... But on pulseaudio.... And I was hearing about pipewire from many different source.. I decided to give it a try... Now my Bluetooth devices works flawlessly....
Thank you for the clear and easy to understand breakdown. I am new to Linux audio, and I'll need to learn a lot because I'm also building a Linux recording setup with Ardour as my centerpiece.
Interesting overview of everything! I'm also using pipewire, which when finally properly set works fine. Only the latency is high for my card when compared to the same device in Windows. For example, sending audio to a hardware fx en receiving the signal back takes about 5.5ms in windows, and the lowest I've seen this in my linux setup is 13.6ms (tested using Bitwig). Using a buffer size of 256 @ 48000. I got the RME HDSPe AIO & RME HDSPe RAYDAT used in conjunction, both excellent cards. Also properly configured everything using a realtime kernel. And for hardware I have a Ryzen 3950X, 16gb ram, running from a 7GB/s SSD. All in all aside from losing a lot of plugins (where I can live with), the latency remains something I'm really bothered by. Anyway thanks!
That was very informative. I came here because I recently bought a Hyperdrum virtual drum kit which allegedly connects via MIDI over Bluetooth to a PC running a DAW. I run Ubuntu Linux and discovered that MIDI over Bluetooth on Linux is troublesome at best and I haven't got it working yet with QTractor. One suggestion was to install Ubuntu Studio which they say comes with all the required audio stuff, eliminating the need to find pieces separately and potentially missing things and generally screwing the system up. Still haven't got it working. But we will persevere. I now have a better understanding of Linux audio and the relationships between all the different parts which hopefully will help. Any suggestions from anyone are most welcome. Thank you. P.S. Can't get it working on a Windows machine either... no that I really want to.
Thanks a lot Tony.. I worked in pulseaudio and I really loved this piece of software... It's just a bit heavy for embedded systems... for eg: AudioFlinger is not as heavy as PA
I got into Linux recently with Arch+KDE Plasma. And I didn't realize that the KDE suite probably came with pipewire packaged with it. I wasn't getting sound at first, so I assumed I didn't have a sound server at all. I installed pulse without noticing pipewire was already there and probably needed some more configuration done to it. Somehow for 3 months those have been running at the same time with pulse taking over all audio from devices, and of all the times a .exe of Dark Souls Remastered might finally have caused a sound issue in the form of muted center sounds (which is the absolute most bizarre error I have seen). It fascinates me. Even then I got Pro Audio stuff to setup eventually too, but each bit is it's own tinker session...
An extremely informative video for Linux users that explains things well but in the name of all that is holy, (and I say this as a native English speaker), could you please speak more slowly! I've used OSS, ALSA and Pulse, and will now try to replace it with Pipewire if Debian allows but think of non-native English speakers. You may have been a rapper in a previous lifetime. :)
I'll second the "great job" comments. This vid is a great place to start for those folks (like me) trying to understand those topics and how they fit together...
Thanks for this video! Been having a ton of audio issues, and this should help me fix them, once I understand it a bit better. Much appreciated! Deserved sub!
Thank you so much for the video. I’m running a development desktop with Ubuntu 20.xx. All my standard devices are working, but I’m trying to test some USB audio output and work with the latency. This really helped with your explanation of the audio servers etc. Thanks again
Though you made some mistakes while you were speaking but then correcting those by providing text was the best thing. Thanks for the video and for the corrections. Thanks bro..
@Tony Tascioglu : As a newbie, trying to familiarise with Linux with a hope to switch over permanently from Windows, I was trying to figure out how to edit video in Linux and make my external mic work to record audio at the same time and I stumble upon this video. I should confess that most of the things went over my head and I couldn’t figure out which one of these programmes should I use to make my mic work. I don’t know what to do. 😢
I tried Linux Mint yesterday, put it on a partition to test it. Everything went well, I like how it looks and feels, but can't get any sound, my rme interface won't work. I tried looking for help and searched for drivers, but everything I came across was too difficult to understand, and what I did understand did not work. I erased the Linux partition and I'm back to Win11. 😂 The 6 hrs I spent trying to get it to work is not worth the headache.
@@lelandgaunt7130 I don’t know what went wrong, but generally the audio should work by default. I am using Fedora and replaced Gnome with KDE Plasma and the default inbuilt sound card worked just fine. I just had some problems with my external mic but ultimately, the internet and UA-cam solved my problem. I am in love with Fedora + KDE. ❤️
Linux Audio is a headache and not easily described how things work out of the box. The documentation isn't helpful as there are so many things, but you explained it very clearly and the main features or fundamental functions of each system.
13:30 I think the rapidly repeating audio pattern is caused by x-runs (failure to update the sound buffer before the deadline) instead of sampling issues. Sampling issues typically introduce noise to the highest frequencies only. As a software developer, Pipewire seems to be pretty well designed except that the design opted for timer based loop to read/write the actual audio hardware instead of interrupt based approach that most audio devices do natively. I would have expected the actual hardware interface to always use interrupt based approach and then use timer based loop for the software interface. Traditionally the way to fix x-runs problems was to run low latency Linux kernel (PREEMPT config or RT patches, in case of Ubuntu you just install package that's called linux-lowlatency-hwe-20.04 where the number matches your distribution and reboot to new kernel) but according to some user reports the above mentioned timer based hardware handling, some audio hardware has now more x-runs with low latency kernel. I don't understand the rationale for this but I have to trust the user reports about it.
@@lawrencedoliveiro9104 Yes, that works in general. However, in case of audio hardware, the soundcard has its internal clock and the system must adapt to it because the clock on the soundcard cannot adapt. In practice, this is implemented by triggering a hardware interrupt activated by the audio hardware. For some reason that I fail to understand, PipeWire doesn't do this but pretends that the system clock is good enough and uses that instead. This despite all the existing hardware actually working as I described above. If system clock and DA converter clock are in perfect sync, this doesn't make any difference but this never happens in practice. I've been writing software for living for a couple of decades now and in my experience API should reflect the hardware environment or you'll see "impendance mismatch" sooner or later as a result of API not actually matching the hardware.
When a Start Jack audio, other sounds of my compter not working. Example, youtube. When started I'm not listen de youtube audio. Anybady with same problem?
Depends if you have anything else running. If you start JACK and it takes the ALSA device for itself (and the ALSA device does not support hardware mixing), then anything else trying to use Pulse to get to the ALSA device might not work. In that case, you could make Pulse output over to the JACK server which passes it off to ALSA. I mention this, since I don't think UA-cam in a browser would have native support for JACK, likely just Pulse.
Absolutely perfect explanation of what I was missing in my knowledge. Thanks for putting the effort into making the video. Also, if you haven't already, shave your head bald. You'll look like a badass.
I had that bluetooth problem with my jabra headset and pulseaudio as well. It required me to keep the pulse audio gui open at all times and then it worked. Eventually rebooted my system and all worked properly again.
Great video! It helped me to organize and understand how things work and why some doesn't work here. I have a Firewire M-Audio 410 here and after moving from pulseaudio + jack to pipewire (0.3.39 on Debian here) I noticed that artifacts you mention at 13:25 when recording from my microphone on the onboard soundcard and the guitar plugged into firewire on ardour. When there is no ardour on the way, just connecting the input of M-Audio to guitarix or rakarrak and then to HDMI, the sound is perfect. It seems to be some resampling issue between the M-Audio and onboard sound card, right? Should I completely remove pulseaudio and jack to have things working well here? I manage the connections using qjackctl, is there some similar alternative? Thanks again for the great video!
There are actually two “versions” of alsa midi, raw and sequencer (that I believe emulates what oss used to do). There are of course simple “bridge” programs between Jack and alsa midi like a2jmidid
Thanks! Looks like BSD support is improving! www.phoronix.com/scan.php?page=news_item&px=PulseAudio-Better-FreeBSD Seems like there are some ports out there, but I haven't tried it personally.
Pipewire started life as a way to handle the gaps left in Wayland for stream capturing with v4l2 and dma-buff, zero copy screen sharing, audio/video sync, etc. They ended up recreating the pulseaudio server as a plugin and reimplementing the JACK interface with a focus on latency. It really is a wonderful project.
Great info! Thanks!!! Have been using Linux for over 15 years, never gave a second thought to how audio works on Linux. After watching this, I uninstalled PulseAudio and installed PipeWire. I have a large living room, and used to get "crackling" when using my Bluetooth headsets. This is no longer the case with PipeWire!! Have only had it installed for an hour though, so will see how it goes.
How are you liking PipeWire so far?
@@Lambda_Ovine I'm loving it! I'm not really an "audiophile", but it does sound better on my setup (no more intermittent crackling). I've converted my laptops over to Pipewire too :) Does everything I need to do, but my needs are very straight forward and simple (i.e. just need audio through Bluetooth speaker). Actually, I WFH, so I video conference too.
Glad to hear that Pipewire bluetooth is working well!
Actually, I noticed the crackle/stutter under PulseAudio on my laptop as wel last yearl, I had tried so many things with Pusleaudio such as forcing SBC (thinking it would use less bandwidth or something), messing with the server delay and buffer times (thinking it was a buffer underrun or something), even turning off WiFi in case both the 2.4 GHz signals are interfering... I guess this did disappear when I switched over and never noticed, neat!
An update to this video, taking into consideration the recent audio technology updates (using ALSA and PIPE WIRE only), would be very welcome. Tank you for this video anyway.
This is exactly what I was looking for- not a tutorial but an explanation of how everything fits together. You rock!!
Thanks! Glad you found it useful!
Yes same here. Thanks Tony!
Thanks for doing such a nice rundown. I have been using PipeWire for a good few months now without any major issues. I had major issues about half a year ago (I know this video was released even before that time). In my experience so far, BT headphones work better with PipeWire than they did with PuldeAudio, though I don't use them that much. An amazing thing is that using PipeWire I can record audio from my BT headset right into Ardour, like any regular audio input. Plus I can work on my music using them in playback only profile. It's pretty amazing.
Yeah, I think the stability and support for Pipewire has greatly improved in the past year. As for recording audio from Bluetooth, make sure you check which sound profiles you're using, since some definitely sound better then others with Bluetooth (SBC-XQ vs SBC for playback for example)!
@@TonyTascioglu Ah, interesting! Yes, the quality in duplex over Bluetooth is usually pretty bad. Having an option to improve that would be great.
YOOOO IT'S UNFA?? HOW IS THIS NOT PINNED??
Really excellent explanation and exactly the kind of patient, thorough guide through linux audio that I never knew I wanted. I recently had some trouble switching my audio output from the hdmi port on my video card to a usb out connected to my headphone amplifier and suddenly I needed to learn about how sound worked.
You seem really knowledgeable and did a great job explaining the full stack of technology powering audio. I would absolutely tune in for more videos like this (even if they take a year to put together :-)).
Thank you!
Hey, thanks a lot for the kind words, glad you enjoyed it!
`pavucontrol` is probably the easiest DE-agnostic way to easily switch inputs and outputs in general or on a per-program basis provided you're using PulseAudio or PipeWire.
I'm a total noob when it comes to linux however If I understand this correctly: Pipewire has the potential to give us pro audio users the ability to route multiple sources to the same and/or different outputs simultaneously. If this is the case, this would be game changing for those looking for advanced audio routing solutions. It would also be much more affordable than using Dante where you would require something like a
CTP DIO88 Analogue to Dante Converter upwards of ($1000/£1000 give or take) to convert your devices and send them where you need to.
Stumbled upon this video while searching for a way to get pipewire to output 5.1 surround audio via S/PDIF. Very informative and while it didn't solve my problem, at least I now know more about PulseAudio and Pipewire. Researching resumes...
Hm, you are probably right. I remember that 5.1 wasn't fully implemented working in Pipewire last year.
While I don't have an SPDIF output or a 5.1 setup to check on my current system, I checked the supported HDMI profiles and it does allow for a 5.1 or 7.1 output (as well as allowing 5.1 digital output on my old SoundBlaster), so it might be possible using HDMI instead if it isn't supported over optical.
I have a 5.1 sound interface connected via USB. All the 6 channels show up for separate connection in qpwgraph, and I was able to route audio to them separately.
This was a really great explainer. I never got how ALSA worked until now. I'm still unsure about Pipewire since it seems so buggy, but hopefully it will be stable enough to try on the next Mint LTS. After all those Ubuntu users test it for us, lol.
Ah yes the Arch/Fedora/SuSe Leap -> Ubuntu/Manjaro -> Debian/CentOS/Mint chain of software propagation...
Jokes aside, thanks for the kind words and glad you found it helpful!
Really appreciate you putting this together. This is a great overview / intro to understanding the ecosystem
Thanks for the kind words! Glad it is helpful!
I've researched a lot about this topic and this was the best explanation I saw on how sound works on linux, thanks a lot for this video.
Glad it helped!
You are an amazing teacher! It is extremely rare to have capacity for knowledge and a great personality. If only grade school was full of interesting people, everything might be different. Thanks for sharing this knowledge strangely two months before I needed it.
Linux Audio is why I daily drive a hackintosh, despite being very keen to switch to Linux. I'm hoping that as Pipewire matures everything gets a lot simpler and easier. Thank you for the video, at least I now have a better grasp of the issues.
Glad you found the video useful!
I remember I thought the same but for OS X audio when I had to use it for a job. Drove me crazy for a bit back then trying to find a way to capture both the system output and microphone input, which I could do on Windows and Linux.
Thank you for keeping this up over the years. It's people like you and content like this that made UA-cam the beautiful place it once was and make me reminisce. Thanks for making me feel nostalgic AND giving me the high level overview I was looking for. Keep on tinkering my friend.
This was very informative. Instead of you going like “yeah use this bc uhh yeah” you actually explained all the options and the pros and cons. Good stuff
great explanation complete from input to output
great work, very informative!
just a footnote: the minimum sampling frequency must always be at least 2x the max frequency being "recorded", otherwise spectral info could be lost, as stated by the Sampling Theorem (Nyquist-Shannon). so even if we can only hear up to ~20kHz we actually need at least ~40kHz in sampling frequency. in reality the distortion caused by the sampling process is almost imperceptible to most people but it exists, and human hearing ranges may vary as well
You are absolutely correct about needing at least double the frequency to capture the original information as per the Nyquist-Shannon sampling theorem - hence most sound cards running at 48 kHz these days.
If you're interested, Chris Montgomery also has a great video exploring and demonstrating the sampling process as well as effects quantizing the input over on Xiph's website: xiph.org/video/vid2.shtml
This video helped me grasp things that I wouldn't have known how to search or troubleshoot. I've been able to solve a problem on my PC thanks to you! Appreciate the info 👍
Everytime I am going to explain this to anyone I am going to remember you !!! Thanks for this consolidated information.
I almote Broke my system while getting my Bluetooth Headset (Microphone) work. But I think I will give up on Bluetooth Headphones atleast for Meetings.
Finally I understand how alsa and pulseaudio interact, and how they relate to OSS! Thanks for the excellent video! Tips for future vids - more b-roll and perhaps slow down a bit
Thanks for the kind words! Early on, I used to try to get b-roll, but then I started to dread making videos (and I couldn't think of much to film for b-roll for a sound server), so I reverted to back simpler filming and editing... As for slowing down, yeah, it's something I need to work on, it kind of 'just happens' when I get a little passionate.
That was a first class system overview; your video really helped me to picture how it all 'bolts together'. Using qpwgraph, I've just set up the laptop microphone -> EasyEffects (so I can use eq, multi-band compression, etc) -> wfview (which then 'feeds' audio to and controls an Icom IC-705 transmitter on my network) and it's all working an absolute treat. All I have to do now is spend a ludicrous amount of time setting up all the EasyEffects components in an attempt to make it sound good(ish). 🙂 Assuming it sounds okay (long story about Icom codecs omitted to save space), the next stage will be to set all this up in a headless Linux box (with a decent audio interface and microphone) and try to emulate an Orban 9105A Optimod-HF using EasyEffects; I might be gone for quite some time! 😀
You have a talent for explaining things clearly. Thanks.
Thanks for the kind words!
Thanks man. It's the only video on Linux audio explained in much detail.desperately required video.
Thanks for the kind words, glad you found it useful!
TNice tutorials is actually a very good and straight forward tutorial. No having ask questions or guess, no over-explaining elents, and brings
Thanks for the kind words!
You are the man!!! You finally helped me understand how all these different processes work on my Linux machine. Your approach to explaining everything was genuinely balanced with simplicity --yet covering all the nuances of the interplay. Bravo!!
Glad I could help!
Thank you! Very well explained. I have a small music studio and Linux has become my OS of choice in the past year. My Linux skills have not yet been up to the task and I have had to use ALSA exclusively. Pipewire is intriguing me and I will give that a shot. I have one small suggestion for you. If you put a foam pop filter over that mic, your audio would be much improved. Have an awesome day!
Wow! I did not expect this to be so informative.
By the end of the video, you have put every concept, you had introduced in the beginning, together so clearly.
Thank you, really.
Thanks for the kind words!
Your voice is so calming I always drift off! Had to reattach this video couple of times
Great to have all those technologies, their pros and cons and the relation between them explained in one video - thank you!
Thanks! Glad you found it useful!
10:03 One reason PulseAudio makes things easier is it basically slots in under the ALSA API. So apps that know about ALSA automatically work with PulseAudio, without noticing the extra PulseAudio layer between them and the actual hardware.
Dude, you cleared a bunch of doubts I had in just 20 minutes! I bought an old Makie Onix-i 16 channel mixer that only works (so they say - I haven't tested) in linux. Also, the UbuntuStudio distro with the low latency driver is wonderful!
A curiosity: in the old days of 100MHz CPUs, mixing audio was heavy for the processors and digital telephone switchers were built with those processors and requried not low-latency but real-time OS and programming. It was so difficult only 8 countries managed to design and produce them.
Worth every second of my life listening to this. Thumbs up for quickly orienting me on one piece of this confusing morass that is Linux audio technology.
Thanks for the kind words, glad you found it helpful!
ALSA can do software mixing. PulseAudio is not required for that. I listened to "podcasts"[1] while playing SNES emulators on Linux before there was PulseAudio. What PulseAudio can do is to give each program a different volume, resample, and route different programs to different output devices. And what PulseAudio actually did when it first arrived was to add such a huge audio latency that gaming and watching videos under Linux became impossible. Well, I had to uninstall PulseAudio to fix it.
[1]: Wasn't called a podcast back then. Was mp3 files of Chaosradio manually downloaded from an FTP server.
I came here to say something like that. Dmix, dsnoop, plug, route and so on defined in asoundrc it rock solid if a bit cumbersome. There is also the possibility to setup ladspa plugins within the asoundrc settings file to e.g. split and filter audio for a satellite+subwoofer system.
Wow! You covered a lot of ground in one video, Tony--my head nearly exploded! Thanks for an informative march through the sordid history of Linux audio, most of which I have battled with at one time. Linux audio can be quite powerful, a Swiss Army Knife with 99 blades; just be careful when using a blade you don't cut yourself on one of the other 98 blades. Cheers!
thanks so much for this overview! there aren't that many beginner-friendly explanations of how linux's audio backend works!!
Thanks for the kind words, glad it was helpful!
I just Probed my machine Apple product: MacBookPro9,2Linux Mint 21.3 Virginia Ubuntu 22.04 jammy Desktop: Xfce 4.18.1 hoping to work in Katana 50 BOSS Tone Studio OMG! Wish me luck! Thank you for a very Streaght to the point instructional loved the pace of the info you covered. I have sndio , PulseAudio, ALSA, and PipeWire all running now JACK is not running. Working on it keep on keeping on! Thank You Very Much ~ Samuel 🙌💻
This is such a good video holy shit! I have been losing my mind over what these software were and briefly explained all their purpose holistically. Thanks!
This actually expained why I'm having so much trouble with setting up VBAN. Now if I could figure out how to work around the problem..
A lot of ppl have over the years commented on the messiness and confusion around linux sound and could there be an easier way to deal with it. I have used alsa from the command line and pulseaudio via a gui. Not used Jack as I don't mix sound. Pipewire I've not heard of prior to watching this video. An instructive video which explains the basic differences between the various pieces of software and how they interact to provide sound capability for Linux which can be a bit overwhelming for Linux newbies. Keep up the good work.
thank you brunnette ed sheeran
I’ve noticed some sound quality issues on my Debian 12 install so I started researching how audio even works on Linux and which sound server to use. So the gist of everything is to use alsa + pipewire, gotcha.
Using PipeWire for almost two years now and I'm happy to announce that the BT device being "there" but not outputting any beep at all is back again :D Usually resolved by restarting the headphones 😩 From a gut feeling it's probably some chip going into energy saving mode after the last stream closed and it may be related to the headphones themselves. They did start to develop weird glitches over the time (especially the ANC feature) so there's that. Anyway, will totally stick with PipeWire. It really changed everything for me. And that's coming from someone who remembers how to echo "magic integers" into OSS device nodes to get sound going for some really odd native Linux PC games back in the days :D
@10:17 , i have just learnt that i could sync all my daws thanks to jack. This was the best take away from this video for me. Mükemmel!!!
:)
Excellent video bro! I been doing Linux Audio off an on for a couple decades now and only learned enough to navigate and utilize these various apps you mentioned without ever really understanding their interactions. Just getting ready to give it a try again since winter is upon us already. That was super cool! Thanks!
Thanks for the kind words!
Keep up the good work!!! There's a lot of information out there.
Few people are able to dive deep into it and explain the way you did!
Regarding a few comments about your hair... Embrace it!!!
If i were you, i'd even hang an Einstein poster in the background.
Awesome people try to help other's succeed. Hater's try to stop other's success!
You're awesome!!! Keep it up Boss!!!
This is a really great explainer on a rather confusing subject.
I love you, so cool how you know so much and are this passionate about the topic, it seemed so confusing it has been giving me headaches since I switched to linux
Hi Tony, thanks for all the info you've put together here. I've been editing and recording for over a decade on mainstream platforms, and only recently have started to delve into the Linux side of audio. All these bits and pieces of info are of great help! :D
Nice. Finally getting around to moving from xorg to wlroots and pulseaudio to pipewire... I'm a bit of a dinosaur haha so the time has come, been, gone and come again. This was really useful for clarifying some assumptions I'd made about... stuff. Cheers!
What about low latency kernels? Does it really help to improve audio quality?
Harika! Sunum için teşekkürler. :)
I use JACK for everything on my main desktop. It's really easy. Most distros provide a package that helps Pulse and JACK to work together.
Great video, transferred to Linux not long ago, and all of that information was messed up in my head. Thanks.
Glad you found it helpful!
Thanks for taking the time to explain this!
The Same Bluetooth problem happened with me... But on pulseaudio.... And I was hearing about pipewire from many different source.. I decided to give it a try... Now my Bluetooth devices works flawlessly....
Man, what a great video. Thanks so much for this.
Finally, ... I have been waiting so long for a explanation of this. Thanks Tony!
Thanks! Glad you found it helpful!
Thank you for the clear and easy to understand breakdown. I am new to Linux audio, and I'll need to learn a lot because I'm also building a Linux recording setup with Ardour as my centerpiece.
Interesting overview of everything!
I'm also using pipewire, which when finally properly set works fine.
Only the latency is high for my card when compared to the same device in Windows.
For example, sending audio to a hardware fx en receiving the signal back takes about 5.5ms in windows, and the lowest I've seen this in my linux setup is 13.6ms (tested using Bitwig).
Using a buffer size of 256 @ 48000.
I got the RME HDSPe AIO & RME HDSPe RAYDAT used in conjunction, both excellent cards.
Also properly configured everything using a realtime kernel.
And for hardware I have a Ryzen 3950X, 16gb ram, running from a 7GB/s SSD.
All in all aside from losing a lot of plugins (where I can live with), the latency remains something I'm really bothered by.
Anyway thanks!
That was very informative. I came here because I recently bought a Hyperdrum virtual drum kit which allegedly connects via MIDI over Bluetooth to a PC running a DAW. I run Ubuntu Linux and discovered that MIDI over Bluetooth on Linux is troublesome at best and I haven't got it working yet with QTractor. One suggestion was to install Ubuntu Studio which they say comes with all the required audio stuff, eliminating the need to find pieces separately and potentially missing things and generally screwing the system up. Still haven't got it working. But we will persevere. I now have a better understanding of Linux audio and the relationships between all the different parts which hopefully will help. Any suggestions from anyone are most welcome. Thank you.
P.S. Can't get it working on a Windows machine either... no that I really want to.
Thanks a lot Tony.. I worked in pulseaudio and I really loved this piece of software... It's just a bit heavy for embedded systems... for eg: AudioFlinger is not as heavy as PA
I got into Linux recently with Arch+KDE Plasma. And I didn't realize that the KDE suite probably came with pipewire packaged with it. I wasn't getting sound at first, so I assumed I didn't have a sound server at all. I installed pulse without noticing pipewire was already there and probably needed some more configuration done to it.
Somehow for 3 months those have been running at the same time with pulse taking over all audio from devices, and of all the times a .exe of Dark Souls Remastered might finally have caused a sound issue in the form of muted center sounds (which is the absolute most bizarre error I have seen).
It fascinates me.
Even then I got Pro Audio stuff to setup eventually too, but each bit is it's own tinker session...
An extremely informative video for Linux users that explains things well but in the name of all that is holy, (and I say this as a native English speaker), could you please speak more slowly! I've used OSS, ALSA and Pulse, and will now try to replace it with Pipewire if Debian allows but think of non-native English speakers. You may have been a rapper in a previous lifetime. :)
I'll second the "great job" comments. This vid is a great place to start for those folks (like me) trying to understand those topics and how they fit together...
Thanks for the kind words, glad it helped!
Man this was a great and comprehensive explanation. And also very fun to watch.
Thanks !
Thanks for the kind words, glad you enjoyed it!
Very good explanation. Thanks Tony.
This video was exactly what i searched for, thank you, it was very informative
Thanks for this video! Been having a ton of audio issues, and this should help me fix them, once I understand it a bit better. Much appreciated! Deserved sub!
Thanks man. I'm strating on linux and your talk opened my mind. Thanks a lot!
Glad I could help!
Nice video, I learn more about audio on Linux but I hace a question, Katya and pabu control are on the same topic?
Thank you so much for the video. I’m running a development desktop with Ubuntu 20.xx. All my standard devices are working, but I’m trying to test some USB audio output and work with the latency. This really helped with your explanation of the audio servers etc. Thanks again
Thanks, glad you found it useful!
Great video! First time seeing your channel, subscribed! I'm definetly the audience for this type of videos.
Thanks for the kind words!
Though you made some mistakes while you were speaking but then correcting those by providing text was the best thing. Thanks for the video and for the corrections. Thanks bro..
thanks man, finally a complete overall look at linux audio systems. keep it up!!
Many thanks for the explainatiom - thorough and easy to follow! 👌
Thanks man!!! Never saw somebody explain that so well
Pipewire has good sound quality, it finally fix my audio problem on linux !
Good to hear!
@Tony Tascioglu : As a newbie, trying to familiarise with Linux with a hope to switch over permanently from Windows, I was trying to figure out how to edit video in Linux and make my external mic work to record audio at the same time and I stumble upon this video. I should confess that most of the things went over my head and I couldn’t figure out which one of these programmes should I use to make my mic work. I don’t know what to do. 😢
I tried Linux Mint yesterday, put it on a partition to test it. Everything went well, I like how it looks and feels, but can't get any sound, my rme interface won't work. I tried looking for help and searched for drivers, but everything I came across was too difficult to understand, and what I did understand did not work.
I erased the Linux partition and I'm back to Win11. 😂 The 6 hrs I spent trying to get it to work is not worth the headache.
@@lelandgaunt7130 I don’t know what went wrong, but generally the audio should work by default. I am using Fedora and replaced Gnome with KDE Plasma and the default inbuilt sound card worked just fine. I just had some problems with my external mic but ultimately, the internet and UA-cam solved my problem. I am in love with Fedora + KDE. ❤️
Thanks for sharing this information. If possible can you please keep making these kind of videos?
Thanks for reminding me of EMU10k1
Linux Audio is a headache and not easily described how things work out of the box. The documentation isn't helpful as there are so many things, but you explained it very clearly and the main features or fundamental functions of each system.
Thanks for the kind words!
Great explanation, thank you for the information you've provided! 😎
This is the best free software Ive seen. Respect.
13:30 I think the rapidly repeating audio pattern is caused by x-runs (failure to update the sound buffer before the deadline) instead of sampling issues. Sampling issues typically introduce noise to the highest frequencies only. As a software developer, Pipewire seems to be pretty well designed except that the design opted for timer based loop to read/write the actual audio hardware instead of interrupt based approach that most audio devices do natively. I would have expected the actual hardware interface to always use interrupt based approach and then use timer based loop for the software interface.
Traditionally the way to fix x-runs problems was to run low latency Linux kernel (PREEMPT config or RT patches, in case of Ubuntu you just install package that's called linux-lowlatency-hwe-20.04 where the number matches your distribution and reboot to new kernel) but according to some user reports the above mentioned timer based hardware handling, some audio hardware has now more x-runs with low latency kernel. I don't understand the rationale for this but I have to trust the user reports about it.
Timer-based operation is how real-time systems work. Tasks have to run to completion within their allocated time slice.
@@lawrencedoliveiro9104 Yes, that works in general. However, in case of audio hardware, the soundcard has its internal clock and the system must adapt to it because the clock on the soundcard cannot adapt. In practice, this is implemented by triggering a hardware interrupt activated by the audio hardware.
For some reason that I fail to understand, PipeWire doesn't do this but pretends that the system clock is good enough and uses that instead. This despite all the existing hardware actually working as I described above.
If system clock and DA converter clock are in perfect sync, this doesn't make any difference but this never happens in practice.
I've been writing software for living for a couple of decades now and in my experience API should reflect the hardware environment or you'll see "impendance mismatch" sooner or later as a result of API not actually matching the hardware.
Interesting! Thanks for the write up, I had not looked into the implementation in Pipewire and how it fully interacts with sound cards, good to know!
This video was super helpful. Thank you a lot for making it!
Thanks for the kind words, glad it was helpful!
When a Start Jack audio, other sounds of my compter not working. Example, youtube. When started I'm not listen de youtube audio. Anybady with same problem?
Depends if you have anything else running. If you start JACK and it takes the ALSA device for itself (and the ALSA device does not support hardware mixing), then anything else trying to use Pulse to get to the ALSA device might not work. In that case, you could make Pulse output over to the JACK server which passes it off to ALSA. I mention this, since I don't think UA-cam in a browser would have native support for JACK, likely just Pulse.
Absolutely perfect explanation of what I was missing in my knowledge. Thanks for putting the effort into making the video. Also, if you haven't already, shave your head bald. You'll look like a badass.
Your hairstyle make me believe in you. 13:20 you also good in rap apparently, love it!
Having discovered Cadence, I'm pleasantly surprised how easy it is to have pulse on Jack. Cadence simplifies it to clicking a couple of buttons.
ABSOLUTELY... EXcellent!!! thanks
I had that bluetooth problem with my jabra headset and pulseaudio as well. It required me to keep the pulse audio gui open at all times and then it worked. Eventually rebooted my system and all worked properly again.
Thank you Tony I really needed a video like this
Very helpful Tony, thanks for your educated perspective on all of this! -Jan
Thanks for the kind words Jan, glad you enjoyed it!
Great video! It helped me to organize and understand how things work and why some doesn't work here. I have a Firewire M-Audio 410 here and after moving from pulseaudio + jack to pipewire (0.3.39 on Debian here) I noticed that artifacts you mention at 13:25 when recording from my microphone on the onboard soundcard and the guitar plugged into firewire on ardour. When there is no ardour on the way, just connecting the input of M-Audio to guitarix or rakarrak and then to HDMI, the sound is perfect.
It seems to be some resampling issue between the M-Audio and onboard sound card, right? Should I completely remove pulseaudio and jack to have things working well here? I manage the connections using qjackctl, is there some similar alternative?
Thanks again for the great video!
Great video, thank you!!
Amazing video, Thank you very much for your effort!
My pleasure! Glad you enjoyed it!
Just as in hardware ( soundcards and samplers) multiplexing degrades soundquality when done from one single DAC
Lovely vid Tony, quite clarifying.
Thanks for the kind words!
8:45 ALSA also does MIDI. But ALSA MIDI is not compatible with JACK MIDI.
There are actually two “versions” of alsa midi, raw and sequencer (that I believe emulates what oss used to do). There are of course simple “bridge” programs between Jack and alsa midi like a2jmidid
Great vid! I'd be interesting in hearing where these technologies are along in the bsd community
Thanks!
Looks like BSD support is improving! www.phoronix.com/scan.php?page=news_item&px=PulseAudio-Better-FreeBSD
Seems like there are some ports out there, but I haven't tried it personally.