Announcement

Collapse
No announcement yet.

PipeWire 1.0 Released For Managing Audio/Video Streams On The Linux Desktop

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Originally posted by Daktyl198 View Post
    PipeWire is definitely an improvement over Pulse, but it still has issues with USB DACs that have both an output and input component. Specifically, it has issues with activating both the output and input at the same time. There are multiple bugs about this on the git repo, one of which I filed myself weeks ago but has 0 comments on it.
    This is one of my big beefs so far with audio management with Linux.

    So far I have been unable to setup audio input and output paths for any devices, USB or otherwise.

    Example: I have a Allen & Heath ZED8i mixer and use the USB link for audio I/O. But when I connect my work audio headset, a Jabra DECT USB cradle to the Linux system, it takes over all microphone and audio out functions.

    I want to be able to patch by a GUI the inputs and audio outputs of each available device.

    So when I am on a video/audio conference call via a Zoom or HVD, I want to patch that Jabra device and assign it to the application.

    When I am listening to hi-def audio via the Allen & Heath mixer and my Sennheiser headphones, I want the audio for my music player to route to the mixer and by extension to my speakers or headphones, not to the Jabra headset.

    As it stands today, I feel like an old AT&T plugboard operator pulling and plugging in the USB cables I need to fit the use case.

    Surely someone out in the ether has an answer. I will gladly pay for such an application, because it would make me more productive as well.

    Comment


    • #32
      Originally posted by edwaleni View Post

      This is one of my big beefs so far with audio management with Linux.

      So far I have been unable to setup audio input and output paths for any devices, USB or otherwise.

      Example: I have a Allen & Heath ZED8i mixer and use the USB link for audio I/O. But when I connect my work audio headset, a Jabra DECT USB cradle to the Linux system, it takes over all microphone and audio out functions.

      I want to be able to patch by a GUI the inputs and audio outputs of each available device.

      So when I am on a video/audio conference call via a Zoom or HVD, I want to patch that Jabra device and assign it to the application.

      When I am listening to hi-def audio via the Allen & Heath mixer and my Sennheiser headphones, I want the audio for my music player to route to the mixer and by extension to my speakers or headphones, not to the Jabra headset.

      As it stands today, I feel like an old AT&T plugboard operator pulling and plugging in the USB cables I need to fit the use case.

      Surely someone out in the ether has an answer. I will gladly pay for such an application, because it would make me more productive as well.
      If I understood correctly, you want the ability to be able to route audio pretty much anywhere to anywhere, similar to what Carla lets one do with JACK? qpwgraph seems to do that for Pipewire, as one example of such an app. Carla seems to work with PW as well, but it's working as a JACK client, so not sure if it's the best way to manage the connections in Pipewire..

      Comment


      • #33
        Originally posted by edwaleni View Post

        This is one of my big beefs so far with audio management with Linux.

        So far I have been unable to setup audio input and output paths for any devices, USB or otherwise.

        Example: I have a Allen & Heath ZED8i mixer and use the USB link for audio I/O. But when I connect my work audio headset, a Jabra DECT USB cradle to the Linux system, it takes over all microphone and audio out functions.

        I want to be able to patch by a GUI the inputs and audio outputs of each available device.

        So when I am on a video/audio conference call via a Zoom or HVD, I want to patch that Jabra device and assign it to the application.

        When I am listening to hi-def audio via the Allen & Heath mixer and my Sennheiser headphones, I want the audio for my music player to route to the mixer and by extension to my speakers or headphones, not to the Jabra headset.

        As it stands today, I feel like an old AT&T plugboard operator pulling and plugging in the USB cables I need to fit the use case.

        Surely someone out in the ether has an answer. I will gladly pay for such an application, because it would make me more productive as well.


        Originally posted by lem79 View Post

        If I understood correctly, you want the ability to be able to route audio pretty much anywhere to anywhere, similar to what Carla lets one do with JACK? qpwgraph seems to do that for Pipewire, as one example of such an app. Carla seems to work with PW as well, but it's working as a JACK client, so not sure if it's the best way to manage the connections in Pipewire..
        Yup, pipewire can do this easily, in fact, you can even just use any jack client perfectly fine. works flawlessly in my experience.

        it's worth noting for some devices you need to enable "pro audio" to show all inputs and outputs. IIRC qpwgraph. I just wish I could export the patchbay view as some kind of wireplumber config

        Comment


        • #34
          Originally posted by Type44Q View Post
          "Look, spotted the delusional guy who lives in his non-scientific bubble."

          I've been accused of many things but that's a first - rest assured I'm at least as aware of all the "theory" involved as you are (not only do I have four years of college-level physics under my belt, my aspie sigint dad taught me electrical theory when I was six).

          "So how high can you hear in frequency huh?​"

          Make that far more aware: as stated by reavertm above, Sampling rate != Sound frequency. The upper range of my hearing (which, incidentally, I have to have tested every two years - and the sensitivity of which always blows the mind of the person testing it) doesn't enter into the discussion.

          I'm the farthest thing from a "deluded wannabe audiophile" - I run twenty-year-old Class A/B amps (the butchered sound of Class D quite literally makes me mildly queasy) through no-name cables powering thrift store speakers - albeit good thrift store speakers like B&W's, Paradigms, DefTechs and Klipsch References (I had the good fortune of living and working in a ski area for the past ten years where Texas oil money regularly donates the unused electronics in their multi-million dollar chalets to charity).

          You probably won't be inclined to acknowledge the fact but we don't all have equally-mediocre nervous systems: not only do I have amazing hearing, my sense of time is far more fine-grained than that of most people (I can entertain a dozen different thoughts just while you're struggling to complete a sentence). I used to be utterly unable to listen to uncompressed 16bit recordings of Upside Down by Diana Ross because of the artifacts; it's only now that I'm able to enjoy (far-smoother) 24bit recordings of it that it's not jarring to my ears.​

          "To me what "sounds good" is something that reproduces the exact input signal it's fed, as close as possible."

          You hit the nail on the head in spite of yourself; the above is all that matters and represents the entirety of the approach I stick to.

          Comment


          • #35
            Originally posted by Veto View Post
            Actually it is not that easy to do. That will require an infinitely steep anti aliasing filter...

            So the better DACs will instead up-sample the signal and apply digital filtering to relax the transition band requirements and thereby the phase-shift of the analog filter.
            I don't think a typical DAC has a AA filter, that is most likely done in software. A ADC on the other hand will probably have some analog filtering to avoid aliasing on the input.

            So if you don't apply AA in your chain you can output SR/2 sinus without problems. Aliasing only happens if you have signals above SR/2 and don't cut them away with a filter. https://en.wikipedia.org/wiki/Aliasing

            Originally posted by Type44Q View Post
            I'm the farthest thing from a "deluded wannabe audiophile"
            But you claim to hear frequencies above 20 kHz and a difference between 16 and 24 bit. While you probably don't have the hardware to be able to hear a difference. You would need a anechoic chamber and an amp that has > 120 db SNR as well as speakers that can output > 120 dB. And regularly listening to sounds over 120 dB would have damaged your hearing, so your hearing threshold would be far below 20 kHz.
            Last edited by Anux; 27 November 2023, 12:35 PM.

            Comment


            • #36
              Originally posted by Weasel View Post
              Yeah, sampling rate is just the upper limit of frequency, and that's literally it.

              Every DAC in the past 30 years or more filters the conversion through capacitors so the result is a perfect sinusoidal analog wave. The only thing higher sampling rate does is enable it to have higher frequency content available. But you, as a human, probably won't even hear above 18k Hz, much less 21k Hz (limit of 44k sampling rate). Highest frequency that can be output by a given sampling rate is half of it.
              But it doesn't mean that it is useless; for a start, you may wish to use pipewire to record ultrasound (relevant for industrial applications, animals that make sounds in this band, etc.). Even when strictly processing human-audible sounds, higher sample rate gives larger headroom for filters, synchronisation, generators and other DSP stuff. Half of the sampling rate (Nyquist freq) is not only the max freq representable, but, because of aliasing, it is also a wall that "bounces back" (folds) all signals with a higher freq deep into low-freq artefacts. With 44kHz sampling you are near this wall, so you need sophisticated DSP, with 192kHz you can just brute force and generally get away with this.

              Anyhow, the pw spell to force 192kHz is

              Code:
              pw-metadata -n settings 0 clock.force-rate 192000

              Comment


              • #37
                Originally posted by mb_q View Post

                But it doesn't mean that it is useless; for a start, you may wish to use pipewire to record ultrasound (relevant for industrial applications, animals that make sounds in this band, etc.). Even when strictly processing human-audible sounds, higher sample rate gives larger headroom for filters, synchronisation, generators and other DSP stuff. Half of the sampling rate (Nyquist freq) is not only the max freq representable, but, because of aliasing, it is also a wall that "bounces back" (folds) all signals with a higher freq deep into low-freq artefacts. With 44kHz sampling you are near this wall, so you need sophisticated DSP, with 192kHz you can just brute force and generally get away with this.

                Anyhow, the pw spell to force 192kHz is

                Code:
                pw-metadata -n settings 0 clock.force-rate 192000
                I never said it's useless. I specifically said that it's useful even for shit DACs that only "sound good" at their natural sampling rate. If that happens to be 192khz, then use it. It doesn't mean it will sound better than 48khz, it will just sound better for that specific DAC because its 48khz is broken and likely adds aliasing (shit DAC, there's tons of them out there, so).

                Comment


                • #38
                  Originally posted by Veto View Post
                  Actually it is not that easy to do. That will require an infinitely steep anti aliasing filter...

                  So the better DACs will instead up-sample the signal and apply digital filtering to relax the transition band requirements and thereby the phase-shift of the analog filter.
                  Yeah, just twice the sampling rate is typical. It's easy because the sampling rate is fixed, and not infinite.

                  Comment


                  • #39
                    Originally posted by mb_q View Post

                    But it doesn't mean that it is useless; for a start, you may wish to use pipewire to record ultrasound (relevant for industrial applications, animals that make sounds in this band, etc.). Even when strictly processing human-audible sounds, higher sample rate gives larger headroom for filters, synchronisation, generators and other DSP stuff. Half of the sampling rate (Nyquist freq) is not only the max freq representable, but, because of aliasing, it is also a wall that "bounces back" (folds) all signals with a higher freq deep into low-freq artefacts. With 44kHz sampling you are near this wall, so you need sophisticated DSP, with 192kHz you can just brute force and generally get away with this.

                    Anyhow, the pw spell to force 192kHz is

                    Code:
                    pw-metadata -n settings 0 clock.force-rate 192000
                    Well, yes, if you want to record and process frequencies that are above half the 'standard' 44.1 KHz sampling rate, you need a higher sampling rate. This is uncontroversial.
                    If you are doing subsequent processing of standard audio sampled at 44.1 KHz, it is useful to have a greater bit-depth to allow fiddling around with multiple tracks without worrying about where the noise floor is, and the where dynamic range peaks are. Which means for recording and audio mixing, it is possible to make a case for 24-bit samples, but there is no gain in having the final output anything other than signed 16-bit. You can keep your masters at 24-bit if you ever want to remaster/remix.

                    Here's another Xiph.org page: Xiph.org: 24/192 Music Downloads...and why they make no sense​

                    Ultrasonic intermodulation distortion is a problem, but also explains why people hear real differences between 'the same' track in 16-bit or 24-bit on the same amp & speaker combination.

                    Just like some people prefer the 'warmer' sound of valve amps (due to the even harmonics distortion), others prefer the distortions generated by their favourite system - but it requires some reasonably sophisticated test gear to show what is going on.

                    Comment


                    • #40
                      Originally posted by Weasel View Post
                      I never said it's useless. I specifically said that it's useful even for shit DACs that only "sound good" at their natural sampling rate. If that happens to be 192khz, then use it. It doesn't mean it will sound better than 48khz, it will just sound better for that specific DAC because its 48khz is broken and likely adds aliasing (shit DAC, there's tons of them out there, so).
                      I often hear about shitty DACs but never heard one, to my knowledge there hasn't been any bad DACs since year 2000, even the cheapest are good enough that most people can't hear a difference.
                      Can you name any specific DAC that is bad?

                      Comment

                      Working...
                      X