Is it me, or is sound really great on FreeBSD?

[Tieks]

Looking through these reddit posts, the difference in quality could be the result of problems with resampling (to bring different audio sources to a common sample rate in order to mix them). The algorithms used to alter digital music are more mathematical than musical. The more you alter digital music, the worse it gets. The loss of spaciousness some people report is a typical result, it could even lead to distortion.

One could rule this factor out by putting the device in bitperfect mode, but I don't know if Alsa/Pulseaudio give you such an option.

 

[cracauer@]

Looking through these reddit posts, the difference in quality could be the result of problems with resampling (to bring different audio sources to a common sample rate in order to mix them). The algorithms used to alter digital music are more mathematical than musical. The more you alter digital music, the worse it gets. The loss of spaciousness some people report is a typical result, it could even lead to distortion.

One could rule this factor out by putting the device in bitperfect mode, but I don't know if Alsa/Pulseaudio give you such an option.

Jackd doesn't do any resampling, it sits with a fixed sample frequency directly on the hardware. Sample rate conversion is done by applications.

 

[Voltaire]

Jackd doesn't do any resampling, it sits with a fixed sample frequency directly on the hardware. Sample rate conversion is done by applications.

You might find some applications do not work nicely with JACK. Whatever you'll notice, many apps require a plugin to work with jack, and sometimes the packages with that plugin just aren't available on the Linux distro you've chosen.

One could rule this factor out by putting the device in bitperfect mode, but I don't know if Alsa/Pulseaudio give you such an option.

PulseAudio has been capable of bit-perfect output for years. The problem is that somehow it sounds different than on FreeBSD. On FreeBSD you have the option to disable vchans. This is also recommended by the documentation for this bitperfect mode. But I think PulseAudio can't do this. As a result, with PulseAudio your sound always goes through a software layer that you cannot switch off, and that distorts your sound. That seems to me the explanation.

 

[ziomario]

Its you and it's totally sane and do you know why ? because with FreeBSD you found the taste for discovery and challenge, which are completely missing in Windows. Windows is not for the users who wants to experiment,try,understanding. Windows is for the users that have no time or will to fail over and over again before to find the solution. At the same time,FreeBSD is not so uselessly complicated like a lot of linux distributions.

 

[ProphetOfDoom]

I know a valid reason why people think vinyl sounds better, although a CD is technically superior. The re-releases on CD from earlier released vinyl records often sounded worse. Especially the ones that featured a SPARS code ADD. Here they took the original analog recording, then used digital equipment to (re)edit and (re)mix. I don't know what equipment that was, but the result always sounded different and seldom good. Many greatest hits compilations suffer from that same problem.
Disclaimer: I don't play vinyl records any more.

Tieks I wasn't aware of the existence of the SPARS codes at all actually but that makes perfect sense. I volunteered at a recording studio back in Winter 2000 and even then every bit of kit they had was 16 bit. Recording from analog tape to 16 bit absolutely crushes and degrades the sound. I think compilations are particularly badly affected also because every song is adjusted to have a similar perceived loudness. But any adjustment post-mastering is bad because mastering adds inaudible high-frequency noise. Re-mastering risks "moving" that noise into the audible range.

I used jack before. But i had after each reboot "rewire" the jack config which was painfull. Maybe i missed something.

Alain De Vos I'm pretty sure the QT frontend for Jack, qjackctl, allows you to save all your connections to be restored later?
Also it's true that Jack doesn't play nicely with a lot of other software but that's sort of an unfair criticism because most people who use Jack will be dedicating their entire system to audio production and definitely won't be trying to watch YouTube videos, playing games, etc. If you've ever tried listening to internet-streamed music on professional speakers (monitors) you'll know why. They're designed to reveal every flaw in the source material and you can literally hear every artefact of the compression algorithm in all its jarring, grating glory.

 

[PMc]

I used jack before. But i had after each reboot "rewire" the jack config which was painfull. Maybe i missed something.

I think so. The patchbay can be written to file, and I think it loads automatically to the last one used when starting the QJackCtl.
The patchbay is in fact phantastic. It's a little monkey that sits in your machine, and whenever you start an application, it immediately plugs the cables to the place where you want them.

Operating the patchbay from the GUI is weird (there is a tutorial somewhere on the web), but the files are XML: not only can they be edited, they could even be generated from automation.

 

[cracauer@]

[...] because most people who use Jack will be dedicating their entire system to audio production and definitely won't be trying to watch YouTube videos,

Pulseaudio on top of jack takes care of all that, starting with web browsers. The interesting part is that you can route the pulseaudio output through jack plugins, too.

Or to different outputs. I have an audio mixer and all "real" jack applications go to one stereo pair and all that pulseaudio nonsense goes to a different stereo pair. Much more convenient if you want to make adjustments just to the casual stuff without messing up real audio.

 

[PMc]

Also it's true that Jack doesn't play nicely with a lot of other software but that's sort of an unfair criticism because most people who use Jack will be dedicating their entire system to audio production and definitely won't be trying to watch YouTube videos, playing games, etc.

I don't think so. Jack was the only thing that did properly work with firefox, when at some time the default way did eat lots of cpu or do other strange things. And Jack does still work properly, with the only exception that once it was secretly upgraded from jack-1 to jack-2, and then nothing did work anymore, and I had to drop OSS-from-ports and go back to the default OSS.
In my understanding, there is a main difference, the difference between things that do work and things that do not work. And only on top of that comes the matter of quality.
For instance, if you happen to burn-in test a server board in the desktop, then there is no sound. That is *not works*. You need some device that can produce sound - and a one-dollar-fifty usb dongle will do the job. That sounds terrible, but if you need to listen to a video on how to disassemble your gadjawoo, you can then do that. So this is *works*.

And in my experience jack usually works.

you've ever tried listening to internet-streamed music on professional speakers (monitors) you'll know why. They're designed to reveal every flaw in the source material and you can literally hear every artefact of the compression algorithm in all its jarring, grating glory.

Haha, I don't have that problem. I still don't have proper speakers.
I intended to buy some, but that reveals itself as a difficult issue. I cannot go to an audio shop, because after a maximum of two minutes there the sales person will be pissed of me and I will be very pissed of the salesperson - because, they want to know what sound I want to hear (and then sell me that); while I want to buy something that produces the sound that is on the media I play. So this just doesn't work out.
Then I got the idea instead to buy speakers from Thomann. But that again is a difficult matter - the price-tags there are very, hm, scaleable - and I am lacking the expertise to understand what I would actually get for the money, i.e how much I should spend.

 

[ProphetOfDoom]

Pulseaudio on top of jack takes care of all that, starting with web browsers. The interesting part is that you can route the pulseaudio output through jack plugins, too.

Or to different outputs. I have an audio mixer and all "real" jack applications go to one stereo pair and all that pulseaudio nonsense goes to a different stereo pair. Much more convenient if you want to make adjustments just to the casual stuff without messing up real audio.

I actually haven't used jack seriously since circa 2010, so maybe my observations only apply to the historical jack. I recall having to do all kinds of kludges and workarounds with jack on Ubuntu including running a program to disable pulse audio whilst jack was running. There was definitely none of the interoperability you describe.
So anyway I eventually got a MacBook and oh my, does it help to have tools designed for the job. Recording music on macOS is just such a smooth ride compared to Linux. I remember saying on this forum that the free software world lacks decent music software and someone cited FFTW. But the thing is no-one has built anything good that leverages FFTW. The best effort so far in terms of effects is the CALF Studio Gear but even they sound a bit sterile and digital.
To return to analog vs digital, on macOS I use the PSP MixPack2 plugins and the Xenon limiter. The MixSaturator and MixPressor can make basically any guitar or vocal sound like it was recorded in 1969. The Xenon makes things literally sound four times as loud. There is simply nothing in the free software world that's even vaguely comparable.

I don't think so. Jack was the only thing that did properly work with firefox, when at some time the default way did eat lots of cpu or do other strange things. And Jack does still work properly, with the only exception that once it was secretly upgraded from jack-1 to jack-2, and then nothing did work anymore, and I had to drop OSS-from-ports and go back to the default OSS.
In my understanding, there is a main difference, the difference between things that do work and things that do not work. And only on top of that comes the matter of quality.
For instance, if you happen to burn-in test a server board in the desktop, then there is no sound. That is *not works*. You need some device that can produce sound - and a one-dollar-fifty usb dongle will do the job. That sounds terrible, but if you need to listen to a video on how to disassemble your gadjawoo, you can then do that. So this is *works*.

And in my experience jack usually works.


Haha, I don't have that problem. I still don't have proper speakers.
I intended to buy some, but that reveals itself as a difficult issue. I cannot go to an audio shop, because after a maximum of two minutes there the sales person will be pissed of me and I will be very pissed of the salesperson - because, they want to know what sound I want to hear (and then sell me that); while I want to buy something that produces the sound that is on the media I play. So this just doesn't work out.
Then I got the idea instead to buy speakers from Thomann. But that again is a difficult matter - the price-tags there are very, hm, scaleable - and I am lacking the expertise to understand what I would actually get for the money, i.e how much I should spend.

I think I may have just recently joined you in the club of "people who have no professional speakers" as someone recently sprayed flash (cleaning spray) on my £750 M-Audio active monitors (he was trying to help). I haven't dared to test them since.
And yeah, salespeople make me start to curse pretty quickly too. I can't really offer advice on your prospective purchase as I'm just a hobbyist not an audiophile. All i can say is that M-Audio speakers are probably too accurate if your aim is to actually enjoy the sound. Some colouration is good.

 

[YYY__]

If your primary day-to-day tool is the operating system, like infrastructure, it makes a lot of sense. Most of the time, the focus is on the applications that run on top of it.

Its you and it's totally sane and do you know why ? because with FreeBSD you found the taste for discovery and challenge, which are completely missing in Windows. Windows is not for the users who wants to experiment,try,understanding. Windows is for the users that have no time or will to fail over and over again before to find the solution. At the same time,FreeBSD is not so uselessly complicated like a lot of linux distributions.

 

[hazy]

FreeBSD audio seems better than Linux or Windows to my ears as well. Bitperfect mode is a delight when it works, but even without it's a substantial improvement.

 

[Voltaire]

I installed Void Linux on an old laptop and wanted to use PipeWire in combination with Jack and mpv to play movies on the TV via HDMI.
I've spent hours trying to get this to work, but I haven't been able to.

PipeWire + Jack + mpv + HDMI doesn't seem to work on Void Linux, although Void is known as possibly the highest quality Linux distro.

With a second attempt I have now succeeded in getting mpv to use the JACK Audio Connection Kit.
You can see the necessary settings for mpv in this screenshot.

Using ALSA+Jack makes it sound better than using PulseAudio. But it still sounds edited compared to playing the same song in mpv on FreeBSD + bitperfect mode + real-time sound settings.

In the song I just played I hear that the voice sounds less rough in ALSA + Jack compared to OSS4 + bitperfect mode + real-time sound. Even FreeBSD out-of-the-box has a rougher more raw sound than ALSA + Jack. Are there devices that can accurately detect and compare the differences in audio output? I'm sure Jack edited the sound in some way.

I can hear it pretty clearly anyway, so I suspect that out-of-the-box Jack is not using bitperfect mode. Is that correct?

 

[recluce]

I continue to be very happy with audio on FreeBSD. Four years ago, I moved my two HTPCs from Xubuntu to FreeBSD because of audio issues that caused pass-through (simple 5.1 AC-3 and DTS) not to work at all. Also, audio lag caused "wow and flutter" effects when the player tried to sync video and audio. Nothing like that on FreeBSD.

I have since upgraded my main audio system to something really nice, a high end 9.1.4 audio system with Dolby Atmos and all the bells and whistles (Emotiva XMC-2 with Emotiva Class D amplifiers, speakers are Axiom for the main five channels, Emotiva for Atmos and Surround channels). FreeBSD (over HDMI on a simple NVidia GT1030 card) handles it all - pass-through works flawlessly for all flavours of Dolby (up to TrueHD with Atmos) and DTS (up to DTS:X). PCM audio even works for unusual formats like SACD, which uses some really uncommon sampling rates. Raw PCM 7.1 surround (some Japanese Music Blu-rays use this) works as well.

 

[bsduck]

Voltaire Did you also compare sound quality on Linux between mpv->jack->alsa and direct mpv->alsa?

Music Blu-rays

Thanks for letting me know this was a thing, I had never heard of it.

 

[recluce]

What I mean is Blu-Rays of concerts and similar. Even though there are some very few that only have static video images.

 

[Voltaire]

Voltaire Did you also compare sound quality on Linux between mpv->jack->alsa and direct mpv->alsa?

I'll give direct mpv->alsa a try when I have time. Most people usually end up using either PulseAudio or JACK. I've used Linux as a daily driver for over 10 years and the way I remember it is that apps like Bitwig really need JACK or else they won't work.

cracauer@ also said the following things.
Real audio applications use jack both on Linux and on FreeBSD. The major obstacle is Chrome, which really likes pulseaudio.
Jackd doesn't do any resampling, it sits with a fixed sample frequency directly on the hardware. Sample rate conversion is done by applications.

I expected that by using JACK I would hear a sound that is competitive. But that was anything but the case. There are a few songs that I know inside out and have actually played many times (on FreeBSD). The voice is much less expressive in ALSA+JACK and it also lacks detail in all frequencies. If the person sings low notes I can tell it's not correct, but for some reason I hear it even more strongly in the parts where the singer switches to falsetto or to extremely high notes. The singer is someone who can sing better/more correctly tonally than 99.9% of the singers, and I've never heard her sing a wrong note, even in live concerts.

When I play those songs on ALSA+JACK and she switches to falsetto and super high notes, it just sounds really out of tune. There are artists who produce music professionally on Linux distributions so I actually think they might not have that much of an eye for detail and maybe they don't listen to artists who can sing the very highest notes flawlessly (live).