I have been lurking here for a while, and I feel there are some things about Linux that is ignored.
First of all I don't wan't to push an operating system down anyones throat, if you like Windows use it, the same goes for Linux. I use both for different things, I see it as the right tool for the job.
One thing that I think some of you guys miss, is that Linux is the DIY of operating systems. If you don't like it, change it, and see if the patches are good enough to be accepted. If you are not into tweaking things at source level, use some mature distribution of Linux or use Windows,
What I'm saying is that, it would, in theory, be possible to gather up a group of audiophiles knowledgeable of programming C, and change the whole audiosystem from the hardware up, of Linux for something that is better.
At the moment I have not looked that much at the code the samples go through in Linux, but in theory there is big potential of screwing things up by rounding thins down and mixing signals at a to low bit resolution. But at least if I wanted to, I could know, while on Windows I have to believe that Microsoft did it right, or resolve to some blackbox optimizer application, that I cannot tell what is doing, because the guy is making money on the thing (which he is of course entitled to like everybode else).
I have an EeePC701 in the kitchen playing music from my server and radio from the Danish Broadcast Service (Danmarks Radio). This is relevant because Android uses the Linux kernel at it's core, like every other Android phone in the world, so at least the kernel is widely spread, since the comming of Android.
But I am trying not to be religious, I use Gentoo (a source based distribution), because I like to fiddle with things, others do not and should use something else, including Windows, MacOSX, ReactOS, Android, and the like.
Martin
First of all I don't wan't to push an operating system down anyones throat, if you like Windows use it, the same goes for Linux. I use both for different things, I see it as the right tool for the job.
One thing that I think some of you guys miss, is that Linux is the DIY of operating systems. If you don't like it, change it, and see if the patches are good enough to be accepted. If you are not into tweaking things at source level, use some mature distribution of Linux or use Windows,
What I'm saying is that, it would, in theory, be possible to gather up a group of audiophiles knowledgeable of programming C, and change the whole audiosystem from the hardware up, of Linux for something that is better.
At the moment I have not looked that much at the code the samples go through in Linux, but in theory there is big potential of screwing things up by rounding thins down and mixing signals at a to low bit resolution. But at least if I wanted to, I could know, while on Windows I have to believe that Microsoft did it right, or resolve to some blackbox optimizer application, that I cannot tell what is doing, because the guy is making money on the thing (which he is of course entitled to like everybode else).
I have an EeePC701 in the kitchen playing music from my server and radio from the Danish Broadcast Service (Danmarks Radio). This is relevant because Android uses the Linux kernel at it's core, like every other Android phone in the world, so at least the kernel is widely spread, since the comming of Android.
But I am trying not to be religious, I use Gentoo (a source based distribution), because I like to fiddle with things, others do not and should use something else, including Windows, MacOSX, ReactOS, Android, and the like.
Martin
Very true, take for example "Audiophile Linux" which was/is some kind of Debian Mint re-spin that appeared a few months ago. In this case another IT-expert, who also claimed to be an audiophile, insisted that for the best SQ in audio playback you must use a real-time kernel with jackd.
When this began to gain traction I challenged people elsewhere on the web to compare it to another standard lightweight Linux install - Lubuntu, using "deadbeef" as the audio app. Surprise, surprise, they could not tell the difference.
Good Morning Eudyptes
You are free to think, believe and use what ever you want. The important thing is that you are happy with your system and that it's working well for you. That's the only thing that matters... so go on, use debian and enjoy it
Well this is not belief in the sense of blind faith but borne of practical experience. I have compared a standard windows XP and windows 7 install using foobar with ASIO or kernel streaming to a standard Linux install using both GUI apps like audacious and deadbeef, and CLI tools aplay and mplayer. I perferred Linux on my system.
With all the hype that surrounded jplay, I tried the free trial. I could hear no difference. The same with fideliser. There was for me no discernible difference in audio quality.
Much has be written about jplay and some ugly arguments have broken out on the web when others have alleged that jplay is a scam and a dispute between jpaly and jriver dragged on for some time including accusations that jplay was using other people's logos on their website without permission. In many cases jplay have had a hard time explaining the rationale for their "optimizations" which to my mind have often read like gobbledygook.
In Linux I put aside any rationale arguments and tried various "optimizatoins" the sort of thing you read about in these refs:
https://sites.google.com/site/computeraudioorg/
?????
I created my own re-spin long before "audiophile linux" appeared with a low-latency kernel, jackd and audacious. All the audio files used for the test on the LiveCD I created, and it was booted with a toRAM option with all drives disconnected. Everything was memory resident. None of this needed any coding, just a little research into how to configure things. And the result was?
I could hear no real difference between this "optimised" setup and a standard Linux install with a full blown desktop environment. All this was done on a bog standard quad core PC. Perhaps I was suffering from expectation bias, as there was little rationale to say audio playback needed all sorts of tricks to get the best out of an OS.
So now I'm pretty relaxed about what I use, and would urge others to be wary of hype.
Hi Eudyptes
I completely agree with the most of what you wrote, no question about this. The only thing i see a bit different is JPLAY. In a Dual-PC setup (StreamerMode) it makes a noticeably nice difference.
I maybe used the wrong wording with "Hype". The fact that Server 2012 and my optimizers gathered almost 100'000 views (jplay.eu, computeraudiophile.com and aktives-hoeren.de) in the last three months was probably the reason why i used that term.
As always, most people are very sceptic if the developer himself is telling something about his solution. And i would do exactly the same to be honest.
The only thing i can offer you is, read the reviews i posted earlier. It will take you about three minutes to read them, not more. Every comment is linked to the original posting on the internet. There are well known Reviewers as well as professional recording studios. The blind tests that have been made so far resulted in a special situation. The users who compared their 5'000-10'000€ network streamers/CD-Players want to sell their components now. Thats the reason why there are no publicly available blind test reports available at this point of time. Once these costy devices are finally sold, i will be able to present the blind test results as well.
But for the moment, all i can offer are the comments and reviews linked to on my Homepage.
Take care,
AudioPhil
I completely agree with the most of what you wrote, no question about this. The only thing i see a bit different is JPLAY. In a Dual-PC setup (StreamerMode) it makes a noticeably nice difference.
I maybe used the wrong wording with "Hype". The fact that Server 2012 and my optimizers gathered almost 100'000 views (jplay.eu, computeraudiophile.com and aktives-hoeren.de) in the last three months was probably the reason why i used that term.
As always, most people are very sceptic if the developer himself is telling something about his solution. And i would do exactly the same to be honest.
The only thing i can offer you is, read the reviews i posted earlier. It will take you about three minutes to read them, not more. Every comment is linked to the original posting on the internet. There are well known Reviewers as well as professional recording studios. The blind tests that have been made so far resulted in a special situation. The users who compared their 5'000-10'000€ network streamers/CD-Players want to sell their components now. Thats the reason why there are no publicly available blind test reports available at this point of time. Once these costy devices are finally sold, i will be able to present the blind test results as well.
But for the moment, all i can offer are the comments and reviews linked to on my Homepage.
Take care,
AudioPhil
Last edited:
Reviews are just as subjective as any other non-blind test reports. For me they have no information value, sorry. I have read several absolutely convinced reports which turned out that the setup change evidently did not produce any config change at all. E.g. modprobing snd-usb-audio module with different nrpacks values while the module was already loaded, i.e. not doing any change, even with values above 20 while there is a hard-coded limit throwing error above 20. In all these cases which did no change whatsoever the author claimed to hear a clear difference. I do not trust non-blind tests at all.
That clearly shows how big the market is for making money with things no one needs when it comes to computer audio.
Using Computeraudiophile as reference is almost a joke. Just look at their forums, many people from the business itself that need people to pay them explain their terrific view of the audio world.
AudioPhil,
You mentioned a dual-pc jplay set up, an addition which appeared this year. Surely, this is just a client server model for audio streaming which has existed for years. MPD, Logitech Media Server or VLC are Linux examples. Paying for two Windows licenses seems twice as crazy to many Linux users. Moreover, paying for an OS which, according to the proponents of “optimisers”. needs extensive tweaking, in fact to the point that in the case of jplay the best is to be heard when the screen goes blank and of course you have no network connection doesn't make it seem lees crazy. Paying for an OS where most of it must be stripped out seems perverse, especially when this “optimization” is not free. If there is so much junk in Windows preventing optimum audio playback, why use it at all?
Whether its jplay or something else, these “black box” optimisations seem to start from a similar premise that audio playback is somehow degraded if the PC is busy doing many things and that the standard memory management is deficient. For perfect timing we need the lowest possible latency and nothing must disturb the audio play back as if it where a linear process. So we have to minimise the number of competing processes running on the system and raise any audio playback to the highest possible priority (or reduce the priority of everything else). In the case of multi-core processor we also have the idea that something is gained if the audio playback process threads are tied to a given core (cpu affinity). God forbid that the CPU frequency and/or operating voltage should change mid track, so you must mess with various power settings in your BIOS (c states etc.) and turn off Intel EST, AMD “cool n quiet” etc, etc. Then there's the whole vexed issue of memory management, and supposedly only “in memory” playback can offer the best SQ together with other optimisations. Not to forget that timers (HPET) and schedulers can get thrown into this heady mix.
Do we really need all this?
So yes, I'm very sceptical. To my mind it's a long time since standard PC audio playback was plagued by pops crackles and distortion which is indicative of over-runs and under-runs. It doesn't take much effort in Windows or Linux to get bit-perfect quality audio playback from your typical PC/laptop when used in conjunction with an external DAC. It's up to the individual listener to decide if there is a quantitative improvement in SQ by using a dedicated audio server device like an ALIX m/board + MPD or a similar Intel ATOM based system which is both fanless and uses solid state storage, and can use decent linear power sources. I can't imagine that the “Community squeeze project “ ever considered Windows 2012 server for their squeezebox touch replacement: Community Squeeze
Considering how flexible Linux is, I still wonder why anyone would want to use Windows server as an OS for a dedicated audio device. Is the Linux audio stack (ALSA or OSS4) so awful? You know, the one that doesn't need a 3rd party driver for UAC2.
Each to his/her own. But I doubt if I'd ever adopt Windows as the OS for a multimedia or dedicated audio computer.
You mentioned a dual-pc jplay set up, an addition which appeared this year. Surely, this is just a client server model for audio streaming which has existed for years. MPD, Logitech Media Server or VLC are Linux examples. Paying for two Windows licenses seems twice as crazy to many Linux users. Moreover, paying for an OS which, according to the proponents of “optimisers”. needs extensive tweaking, in fact to the point that in the case of jplay the best is to be heard when the screen goes blank and of course you have no network connection doesn't make it seem lees crazy. Paying for an OS where most of it must be stripped out seems perverse, especially when this “optimization” is not free. If there is so much junk in Windows preventing optimum audio playback, why use it at all?
Whether its jplay or something else, these “black box” optimisations seem to start from a similar premise that audio playback is somehow degraded if the PC is busy doing many things and that the standard memory management is deficient. For perfect timing we need the lowest possible latency and nothing must disturb the audio play back as if it where a linear process. So we have to minimise the number of competing processes running on the system and raise any audio playback to the highest possible priority (or reduce the priority of everything else). In the case of multi-core processor we also have the idea that something is gained if the audio playback process threads are tied to a given core (cpu affinity). God forbid that the CPU frequency and/or operating voltage should change mid track, so you must mess with various power settings in your BIOS (c states etc.) and turn off Intel EST, AMD “cool n quiet” etc, etc. Then there's the whole vexed issue of memory management, and supposedly only “in memory” playback can offer the best SQ together with other optimisations. Not to forget that timers (HPET) and schedulers can get thrown into this heady mix.
Do we really need all this?
So yes, I'm very sceptical. To my mind it's a long time since standard PC audio playback was plagued by pops crackles and distortion which is indicative of over-runs and under-runs. It doesn't take much effort in Windows or Linux to get bit-perfect quality audio playback from your typical PC/laptop when used in conjunction with an external DAC. It's up to the individual listener to decide if there is a quantitative improvement in SQ by using a dedicated audio server device like an ALIX m/board + MPD or a similar Intel ATOM based system which is both fanless and uses solid state storage, and can use decent linear power sources. I can't imagine that the “Community squeeze project “ ever considered Windows 2012 server for their squeezebox touch replacement: Community Squeeze
Considering how flexible Linux is, I still wonder why anyone would want to use Windows server as an OS for a dedicated audio device. Is the Linux audio stack (ALSA or OSS4) so awful? You know, the one that doesn't need a 3rd party driver for UAC2.
Each to his/her own. But I doubt if I'd ever adopt Windows as the OS for a multimedia or dedicated audio computer.
Last edited:
Yeah, funny site. They're very sensitive to any post questioning their sponsor's products. Although some claims & questions over their forum are too dumb to be believed, I was surprised when they published article on building Beagleblack music server. Usually this kind of article won't have any chance over there as it will threatening their sponsor.
Well my office sound system (Lenovo T61 attached to a Music Streamer USB DAC and an external 5400rpm 1TB USB drive) could not play 192K or 96K using Ubuntu without crackles.
I installed the linux Mint (ap-linux) distro and things play just fine now. So some optimization seems to have been well designed for audio playback.
I am not using Jack any more as it seems going straight to ALSA is more "bit perfect" than using Jack. Bit perfecter, is probably the right term.
I installed the linux Mint (ap-linux) distro and things play just fine now. So some optimization seems to have been well designed for audio playback.
I am not using Jack any more as it seems going straight to ALSA is more "bit perfect" than using Jack. Bit perfecter, is probably the right term.
Could you please post the output of uname -a on both systems?
And how do you play the files.
KR
I agree with phofman, other than to say that this sounds like a simple configuration problem. with "pusleaudo" a possilbe culprit.
There is no such thing as " linux Mint (ap-linux) distro", either you installed Linux Mint or "audiophile Linux". If the later, then "ap-linux" is not a distro. Last time I looked it had no repositories if its own. You will have to rely on its creator(s) to publish kernel updates etc. otherwise is uses debian mint repos.
It doesn't contain any special "optimizations" of the type mentioned, discussed or referred to in this thread. For memory, the only thing different from a standard Linux install is a RT kernel and some audio apps, like jackd, being preloaded. I can't remember what "ap-linux" does re: "Pulseaudio".
It was using fluxbox as a lightweight window manager, which could be install on most distros if you wanted. Apart from the the fact that an RT kernel is unnecessary for audio playback, it can be problematic if you installed this in a system where you would expect to use proprietary graphics drivers.
Incidentally, there many linux distros where either "pulseaudio" is not installed at all, or can simply switched off (Ubuntu is not one of them) and that have what you might call a "low-latency" kernel which is suited to multimedia apps.
I'm not going to reinstall Ubuntu. But Mint reports:
Linux t61-audio 3.2.32-rt48 $1 SMP PREEMPT RT Wed Jun 19 02:03:46 CEST 2013 i686 GNU/Linux
I use Clementine with preference set to:
GStreamer audio engine: Output plugin -> Audio sink (ALSA)
Output device hw:1
Buffer duration:1000ms
(The last item buffer duration I had at the default 4000 and have been decreasing thinking that disk reads would take place more often and things like parking the drive head would occur less frequently - but am still experimenting).
Again a Hitachi 5400 RPM Sata drive in an external bay connected via USB 3.0 cable.
The ap-linux folks advise removing pulse audio. My understanding is that if Clementine is using Audio sink ALSA then pulseaudio is not involved. I can try removing pulseaudio.
It appears ap-linux is a Mint distro. I am not upgrading as I assume if the kernel is modified then the various priority settings given to audio applications will be overwritten.
Clementine is not an audio app I use, but did you enter anything in the "output device" field on the playback preferences tab? Clementine is not lightweight is QT based (not 1st choice to install on a GTK based desktop) and uses the gstreamer back end.
My own preference is for "deadbeef", MPD with sonata or ncmpcpp, and Logitech Media Server + squeezelite.
AP-Liux has a FAQ on removing "Pulseaudio". Why is was ever preinstalled I don't know. The website says: "Pulse is just there so you can listen other things like Youtube videos and such. Well I've never had "pulse audio" on any Linux install I've used and guess what I can listen to vids without a problem.
Certainly when you install "ap-linux" you see Linux mint installer screens because, as I said before, it's just a Linux Mint re-spin. You wont find it on the main Mint website. AP-Linux is not a standalone distro.
I'm sure if it was called Fred's Linux you would have ignored it, having "audiophile" in the title is a nice hook. I don't have the time and energy right now to give my critic of it, but suffice to say I think it's wrong headed and misleading, especially for those new to Linux, to maintain you need a RT kernel jackd etc. for quality bit-perfect audio playback in Linux. You don't.
As to any kernel updates the creator(s) may publish. In Linux you do not update the Kernel, but simply install new versions and then remove old unneeded kernels. This should have no effect on "various priority settings given to audio applications" , because there aren't any other than any changes that may have been made to "/etc/security/limits.conf" to grant RT acccess to "audio group" . ( See this ref for eg https://help.ubuntu.com/community/UbuntuStudio/GettingStarted)
You did not say exactly which HRT music streamer you were using. Or whether it is an asynch USB device and UAC1 or UAC2. I vaguely remember USB audio class 1 devices that were asynch have been a problem in Linux. A USB DAC that needs proprietary drivers is not necessarily Linux compatible. Anything XMOS based and UAC2 compliant should work OK. Not all Tenor chips work in Linux and I think Cmedia 6631 might still be work in progress.
My own preference is for "deadbeef", MPD with sonata or ncmpcpp, and Logitech Media Server + squeezelite.
AP-Liux has a FAQ on removing "Pulseaudio". Why is was ever preinstalled I don't know. The website says: "Pulse is just there so you can listen other things like Youtube videos and such. Well I've never had "pulse audio" on any Linux install I've used and guess what I can listen to vids without a problem.
Certainly when you install "ap-linux" you see Linux mint installer screens because, as I said before, it's just a Linux Mint re-spin. You wont find it on the main Mint website. AP-Linux is not a standalone distro.
I'm sure if it was called Fred's Linux you would have ignored it, having "audiophile" in the title is a nice hook. I don't have the time and energy right now to give my critic of it, but suffice to say I think it's wrong headed and misleading, especially for those new to Linux, to maintain you need a RT kernel jackd etc. for quality bit-perfect audio playback in Linux. You don't.
As to any kernel updates the creator(s) may publish. In Linux you do not update the Kernel, but simply install new versions and then remove old unneeded kernels. This should have no effect on "various priority settings given to audio applications" , because there aren't any other than any changes that may have been made to "/etc/security/limits.conf" to grant RT acccess to "audio group" . ( See this ref for eg https://help.ubuntu.com/community/UbuntuStudio/GettingStarted)
You did not say exactly which HRT music streamer you were using. Or whether it is an asynch USB device and UAC1 or UAC2. I vaguely remember USB audio class 1 devices that were asynch have been a problem in Linux. A USB DAC that needs proprietary drivers is not necessarily Linux compatible. Anything XMOS based and UAC2 compliant should work OK. Not all Tenor chips work in Linux and I think Cmedia 6631 might still be work in progress.
I have a terabyte and will have more of music and I find the simple interfaces without playlists etc are difficult. I will try using MPD eventually.
It is the Music Streamer II (not +) and so far it seems to work well. I believe the choppiness I hear is due to poor disk drive performance, and I'm working on improving that. Perhaps a better SATA to USB converter.
Clementine has a problem that if the system is brought down poorly (forced power off) the database gets corrupted and then the next time you bring the system up, everything freezes when clementine barfs on the database.
So will definitely go for the client/server model here with MPD as soon as I figure it out.
Fred's linux would be from the Flinstone age right? And definitely not audiophile quality. No thanks.
It is the Music Streamer II (not +) and so far it seems to work well. I believe the choppiness I hear is due to poor disk drive performance, and I'm working on improving that. Perhaps a better SATA to USB converter.
Clementine has a problem that if the system is brought down poorly (forced power off) the database gets corrupted and then the next time you bring the system up, everything freezes when clementine barfs on the database.
So will definitely go for the client/server model here with MPD as soon as I figure it out.
Fred's linux would be from the Flinstone age right? And definitely not audiophile quality. No thanks.
In my opinion you have way to much variables in your hands.
My usual approach with linux is start simple and go for more complicated stuff. In this way you will learn a lot and you will be sure what does what.
For example:
Choose your distribution and maturity stage: safe = current stable release ; cutting edge = development release. I use debian because of apt (package management program). For my other project I have used gentoo development distribution but it became a pain to maintain the config files, on the other hand everything was compiled from scratch with the proper cpu flags.
I'm starting with a minimal system no X, no aditional daemons just skeleton system and ssh server for remote access (I'm using headless machines , even the mpd server) and once it is in place I use apt to install needed packages and if it is the case optional/recommended packages.
Afterwards on another machine I compile the latest git versions of the daemon/program I like (eg: mpd, sox).
This way I know when something added on top of a working system is messing things up.
The drawback of this approach is that you don't have any fancy graphical interface and you should be friendly with a command text editor and linux file system hierarchy (just to know where the conficuration files are /etc and the system logs resides /var/log and /proc /system is usefull too)
When the system is in a quite stable configuration you go and recompile your kernel
This is totally the opposite way to tune a windows system (mostly blindly by trial and error)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.