[Maxpain]

Hello to all,

I recently decided to make a full turn to Linux (Ubuntu) for reasons of security but I also like the idea and style of Linux machines.

I also care about music and audio quality a lot and I don;t want to miss out on this aspect since I can also dual boot and have it both. I listen my music 99% from the top tiers of Spotify and Tidal.

So I installed Ubuntu...I did my research I removed Pulse Audio completely...and I configured ALSA to use my DAC (micro idsd silver) with no re-sampling just plug in to hw:1,0.

In Windows 10 I listen to Tidal for example with the Desktop app using the direct option meaning there is no volume control involved (apparently) and it also does not use windows mixer.

In Ubuntu there is not a Tidal desktop app and I have to listen through Google Chrome and Tidal's web player.

Well at least to my ears sound in Windows and Ubuntu is very different at the least. I can't explain why...I think that in Ubuntu sound is very open but also sounds kinda grainy and less detailed. That is subjective though the fact is that it sounds fairly different to the point that I am thinking that ALSA colors the sounds intentionally.

Is there anyone that have tried both? or its just me having used to the windows sound for so long?

I am opening this thread for general opinions and experiences with this matter.

Cheers.

 

[xeizo]

Ubuntu by default goes through Pulseaudio and with very low quality set for the SRC. It's aimed at working well with low performance computers. It should sound grainy, as it is grainy.

You can edit the settings file for Pulseaudio, pulsaudio.cfg, to change default bit resolution, default sample rate and quality level of resampling. You can also change the SRC to another better one, like SoX, it's possible to set in pulseaudio.cfg. Additionally you can edit alsa.cfg to your liking as well. It's possible to improve the default SQ a lot this way.

For best SQ one needs to remove Pulseaudio and run on ALSA directly, but it's unpractical as you can only run one stream at a time that way.

When eveything set for bitperfect, Linux and Windows should sound identical.

Easiest sound improvement in Windows is to run sound from a particular app in Exclusive Mode, makes a big difference. The same thing in Linux is to use only ALSA.

 

[Maxpain]

I have removed and purged pulse audio and I am running ALSA alone. In windows I only have the choice to play through Tidal and Spotify desktop apps so nothing I can do there.

The thing is that I do observe a difference in sound quality between Linux and windows 10 even with bit perfect playback with foobar and with strawberry media player for example.

Pulse Audio with best settings (avoid re sampling, avoid mixing etc) is still worse sounding than ALSA.

 

[gimmeheadroom]

The best Linux distro for audio is no Linux distro. As much as I hate Windows and as much as I don't like Linux they each have their place. Windows for gaming, audio, and video. As far as I know nobody makes dedicated DAC drivers for Linux and it's just impossible to tell how many layers of crap are between you and your music. And I have not found a media player for Linux that can support DSD or SACD ISO or MQA. Sometimes mplayer or VLC with with DSD and SACD but usually not very well. Linux is not designed for low latency audio and maybe ASIO on Windows is. So I'm just not spending any time on it. I have Windows boxes that are only used as media players.

I don't know whether you could install Windows on VirtualBox or something similar and have it talk directly to your USB devices but if you can, that would be an option if you don't want a physical Windows box.

 

[Maxpain]

I know that there are no dedicated Linux Drivers for lots of DACS out there. I am lucky cause ifi audio supports linux with generic usb drivers...its plug n play.

There are media players for linux that can output directly to your dac but no dsd or mqa I know.

As far as low latency concerns I don't know if it matters only for playback music but there is real time kernel...there are distros like volumio, aplinux and such but they are dedicated for audio alone and I want to browse the internet...read shpreadshits and do other stuff as I listen to my music. In the end those distros also use ALSA so i bet the sound quality would be the same.

I cant say for sure that in linux audio is worse than windows 10 (even with windows mixer involved) but its different for sure. What I am talking about is that it should be the same when you play music with ASIO in Windows 10 and with ALSA in Ubuntu...cause in theory they are both talking directly to the dac...but its not the same sound at all..

 

[gimmeheadroom]

I can't explain it, so I just avoid it. I think there is no one-size fits all desktop solution so I have different hardware, OS, and apps to do what I want.

 

[Whazzzup]

Antipodes audio servers use Linux. To me that’s the story, although I am no tech

 

[gimmeheadroom]

Servers and clients are a totally different issue. Anything can serve music or streams of data as well as anything else.

I'm serving music on my LAN via FreeBSD with ZFS. Before that I was using OpenBSD.

 

[xeizo]

You could try a real time kernel which has a higher tick rate, sound distros always uses real time kernels.

I've mostly used Arch Linux and while I haven't compared directly I was happy with the SQ after some tweaking.

The only Linux I use now is Clear Linux which is made by Intel, it has the highest performance overall of all distros I know of. Even on AMD hardware. For sound, it's only Windows for me, I have never liked Linux for audio production. Too much configuration before even being able to do a project.

 

[Maxpain]

I know that dedicated distros use real time kernels and they should sound optimal...maybe I will Volumio a try but never the less I want to browse the web while listening to music. Those distros does no support browsers.

I don't mind configuring and tweaking I kinda like the exploration.

I also don't do any music production just playback (listening). What I want is to know that I am not giving up any sound quality for using Linux for Audio. I am not ready to sacrifice sound quality for using exclusively Linux instead of windows.

 

[Roseval]

It is a long time ago I used Linux for audio hence it is all a bit dated: http://www.thewelltemperedcomputer.com/Linux/Linux.htm

Both Win and Linux can be configured for bit perfect audio ( no resampling, no dither, etc).
I don't see any reason why an OS should produce a different "sound" if you effectively bypass all the OS audio by sending the audio straight to the audio device.
There are theories stating that the overhead might differ hence different OS my generate different patterns of EMI/RFI, ripple on the ground plane etc.
I have never seen any measurement or a well conducted blind listening test proving this point.

My personal experience is that it is very easy to hear differences, to easy. This I attribute to our perception. The moment you see anything chancing, you will hear that.
That is how our perception works, in my perception of course

 

[Maxpain]

Yes of course...and lets not forget about placebo effect. It is possible but when we are talking about OS's and what do they do to the audio things are not so subjective.

I mean whether Ubuntu has a layer of prepossessing audio or not. What I want to find out is whether I get true bit perfect outcome when I use only ALSA speaking directly to the DAC (hw:1,0 for example).

Now my personal opinion is that Linux is a very good platform for sound cause you can do whatever you want providing you have the technical skills and of course providing there are drivers. For example in windows you have to use a program lets say foobar with asio to talk directly to the dac. In Linux you just have to make a text file and place it in the right directory. I am talking about the potential there...

 

[Roseval]

That mimics my experience.
At interface value, not much different from XP but if you want to tweak you indeed have to edit a config files manually and know what parameters to supply.
Are you familiar with this one? https://www.head-fi.org/threads/bit-perfect-audio-from-linux.561961/

As far as hw: is concerned.
This is what I call bit perfect by configuration.
It probably works correct but if you want to make sure you have to record the output and compare it with the original file.
Sometimes somewhere (might even be the audio device) some resampling or dithering might be going on.
A simple test can be to send something to the audio device you know for sure it doens't support e.g. a 32 bit or a 384 kHz sample rate
If it still plays something is intervening...

 

[gimmeheadroom]

I don't see any reason why an OS should produce a different "sound" if you effectively bypass all the OS audio by sending the audio straight to the audio device.

https://en.wikipedia.org/wiki/Begging_the_question

 

[Maxpain]

As everyone at the internet says the way to output bit perfect audio from Linux is to use ALSA instead of pulse audio and configure the asound.conf file to not use dmix and talk directly to the dac. That's ok by me I have done it. The problem is that Linux and ALSA in particular have bit problem with Spotify and Tidal. Spotify desktop app (snap spotify) does not support ALSA and Tidal does not have a desktop app at all for Linux.

So I either have to play with lowest quality in spotify web player or just use Windows instead. That's a bummer for me that I want to use just Linux but I have to switch to windows for audio listening.