Linux vs. Windows audio quality - please help

Software
1

Now that color management is solved for me (at least on Linux), time to move to audio. It’s going to be a longer post, so please bear with me.

I noticed certain differences (both positive and negative) playing my favorite EDM songs on my Asus mini PC running Ubuntu Studio at the moment. But before I continue, I think it’s important to mention a few things…

  • I listened to the songs from the same source (YT Music Premium), on the same headphones (HyperX Cloud Alpha)
  • So far, I didn’t compare the songs back-to-back, but I think I know them well enough to tell that something’s definitely off
  • no audio enhancements are enabled on either machine
  • Windows - Realtek drivers / Ubuntu - Pipewire
  • the difference is probably heavily influenced by the sound cards of each machine
  • However, I used to encounter the same problems (more on that below) on my laptop back when it was running Linux

Firstly, compared to Windows, I noticed significant improvement in details… like I was saying to myself “…has that been in the song the whole time?”. Many subtle details that I couldn’t hear before were now obvious. I LOVE that! Some songs feel almost as if I re-discovered them.

But also… I can’t get rid of the feeling that on Linux the songs sound kind of… flat? I’ll do my best to explain in my own words, as so far I lack the knowledge of audio terms… sorry

When there is not much ‘going on’ in the song, it’s ok (like just the piano playing, etc.), but when more elements join (more ‘instruments’, many more details and melodies), it feels like the song goes quieter and sounds flat, like I need to raise the volume. I don’t remember that on Windows or Android phone, but I remember something similar happening back when my Dell laptop ran Ubuntu a few years ago.

This all makes me think that something is wrong (?) about my Linux audio setups and I since it doesn’t happen to all songs, I can’t tell if I’m just over-analyzing or what’s going on. Sometimes even skipping trough the song feels like it’s suddenly kind of okay…

So I’d like to ask you all for help:

  • Is this normal? Or am I going crazy…
  • Can I do something about it?
  • What should I consider to be the benchmark of quality? Any way to test?

I’d love to try EDM production myself, so I want to sort this out before I start.

Oh, and one more thing… This is what I’m running on Linux

image
image542×274 20.9 KB

2

pipewire pipeline defaults to 48kbit, while windows resamples to 44kbit iirc.
android also resamples by default. some music player apps support exclusive mode where they can switch the audio devices to even higher sampling rates depending on the input material.

also if you want to do audio stuff on linux follow https://librearts.org

and I can recommend Unfa on YT :slight_smile:
you will find out you dont need paid software to get started.

also state of azure has some nice synth tracks with vcvrack which is opensource. Vital seems fun too. You can pay for it to support the dev but you dont have to. there are forks of both tools, but i still recommend using the original and supporting their devs. For some good arguments on why see FOSDEM 2025 - The Road To Mainstream Matrix.

also with pipewire and the builtin jack server the whole audio production stack became sooooooo much easier and more fun.

1 Like
3

I wouldn’t pay for a music production software when I have no idea if I’m even going to find it entertaining. I found Tracktion Waveform Free, it is said to be quite feature-rich, although I have zero experience when it comes to music in general, so I might need something simpler…

Will check, thanks!

I’ll be honest, I’m not sure what I’m actually running, I hope I understand that I run basically Pipewire? IDK man, I added a screenshot, please correct me if I’m wrong

4

you are using the pulseaudio compat mode from pipewire. depending on your distribution there might be another package that enables jack support.

you want pipewire most of the time.

There is Ardour, Bitwig, Zrythm just to name a few. Unfa will show you how use them :slight_smile:

2 Likes
5

What should I do? Any tutorial to get rid of Pulse?

6

you are using pipewire. pay attention for the stuff in ()

1 Like
7

Oh, right, I got it now. That would explain why I was missing /etc/pulse/default.pa to make edits and prevent the audio from suspending and making the crack/pop noises after a few seconds of silence. I think it’s fixed now, I found a guy on Reddit mention editing /usr/share/wireplumber/main.lua.d/90-enable-all.lua and comment -- load_script("suspend-node.lua")

What about the quality differences? Am I just making stuff up or is there some truth to what I’m experiencing?

8

YouTube has this setting “stable volume” – maybe it has to set for each browser. For music, I’d set it off. Spotify also has some kind of volume normalisation.

Do you also get the problem with Minitube?

9

image
image736×137 9.21 KB

Never heard of that, I could try
EDIT: the same problem occurs

10

why not test some local files to eliminate the browser, youtube, and all that stuff. If it sounds good locally, then the problem is browser, yt music, or something like that and not pipewire/subsystem type of stuff.

1 Like
11

I don’t really have anything locally that I could use for comparison… I know a specific song where I can tell. Is downloading the song off YouTube the same as playing it in a browser? (using those download services, I don’t use them, but just for testing)

12

Tutorial to get rid of Pulse Audio: rename the executable.

eg /usr/bin/pulseaudo to /usr/bin/pulseaudoX.

But this presupposes that you have something set up that will work without it — otherwise you just completely broke your audio. Which is part of why rename it rather than delete it.

I can’t remember how I got to where I am today, the story started with a lot of PCI Latency problems, switching to Linux, and, for some reason, buying a firewire audio interface. Which would only work with JACK. And JACK was a huge pain to get working, until I found something that made it easy: the KXstudio distro. It also involved realtime and lowlatency kernels, which don’t seem to be necessary any longer.

I now use Linux Mint. But I install the basic JACK stuff from KXstudios It also includes some stuff which looks after your process priorities so that audio gets treated as actually important

Is everything now fine and easy? Nope. But My hearing is just not good enough for me to make it a major project like back then. But do I recognise the symptoms you describe? Absolutely.

I think some audiophiles might describe it as a for of “digititis.” It is like the music has lost its life. The toe no longer taps; one feels like turning it of rather than looking forward to the next track.

And I do think this is real, not some form of audiophool bias. Just because I can’t explain it doesn’t mean there isn’t technical fact at the bottom of it. My idea is that it has to do with process/device priorities. I also don’t think that variable-speed CPU helps: slow, fast or medium is less important than fixed.

First Aid, when your music sounds dull: reboot your machine. Placebo? Of course it could be, but I favour the process-priority/etc explanation.

Sorry I can’t be more specifically useful. Fifteen years ago, I might have been. My source, then, for all things Linux-audio was the linuxmusicians forum.

(I have not met pipewire as yet. Sometimes I think I’ll have a go with it. But honestly I don’t have the enthusiasm to rebuild my audio system: I’d rather put what’s left of my mental energy into darkatable!)

1 Like
13

You should be able to download some free FLAC or other audiophile test files to use for comparison…

2 Likes
14

I feel like one of the reasons I feel this is the general bass heaviness of audio on Windows. It feels ‘punchier’…? Adding bass on Linux helps a bit. Not completely, but hearing details with less aggressive bass so far feels reasonable to me. I’ll need some time to test and settle things in my mind.

No problem.
I actually compared a song back-to-back and the drop in loudness was pretty much the same on both systems, so it may have been a placebo and related to what I said above

15

One possible reason for the discrepancy could be the different sound cards in relation to the headphones.

The headphone amplifiers built into sound cards are often not of good quality. The 60 ohms of the HyperX Cloud Alpha can be a challenge for some sound cards.

The ones built into audio interfaces are somewhat better – I can recommend the Focusrite Scarlett 2i2 3rd Gen, for example. (There’s a video on YouTube from unfa showing how to get it running under Linux.)

This would also give you the opportunity to test the same interface under Windows and Linux, eliminating the possibility of “different sound cards” being the source of the problem. It can easily be switched over via USB.

The sound is even better with a headphone amplifier (connected to the audio interface) – although these devices are completely out of fashion.

I’m not familiar with the HyperX Cloud Alpha, but it’s advertised as a gaming headset. It would probably be advisable to use studio headphones instead.

Otherwise, I remember the sound on my Windows laptop sounding exceptionally good, too. But that was actually due to the sound enhancement enabled by default.
Are you sure everything is disabled here? On both sides?

When making such comparisons, I would definitely make sure the conditions are the same.
For example, recording this song from YouTube with Audacity and then playing the file on both platforms with Audacity.

But isn’t this platform about pixels, not beats?

16

you can also load audio files into image editors.

what is your point?

17

We like to help each other… we have lots of smart and experienced people that know way more than photography, including quite a few audiophiles, as I’ve recently learned.

Isn’t helping other people good? That’s what we aim to do here. We mostly focus on photo things… but if we can help other place, we should.

2 Likes
18

Yes, it is, but it’s the best headphones I’ve ever had…

In Windows I disabled ‘audio enhancements’ somewhere deep in the settings, which helped the quality.
In Linux there was nothing running at the time, now I got EasyEffects installed with a few minor effects applied.

On Asus GR8 I can hear sooo much more details where I couldn’t hear them before, especially certain songs (I can give examples if need be) and overall everything sounds so much… Cleaner, polished? I don’t know how to describe it. This seems to be especially true for heavy dubstep songs, but very melodic ones as well

19

First I would try to use the same output for both windows and linux. A very inexpensive way to do this is to invest in a chi-fi dongle with the cx31993 chip, which you can find for under 10 euro on aliexpress or similar shops. You can also get a JCALLY JM-12(10€) or JM-20 (20€).

These dongles are very transparent and should have enough juice to power your headphones. I would not get into audio interfaces or headphone amps as of now, it’s unjustified and a waste of money for the headphones and use case you have. All things considered the dongle will provide a better experience than the bad dacs/amps found on lower end interfaces(focusrite, behringer, etc). People who doubt this can go look up measurements of those dongles, especially the JM-20 :slight_smile:

After you have the same source it’s a question of tracking down the problem. Probably what’s happening is that something is applying normalization, compression or EQ in windows and that’s where the issue stems from. I highly doubt it’s resampling…

3 Likes
20

Yeah, way too early for that. If I happen to really like EDM production then maaaybe…

Yep, I could try that. I found something at around 16€ with shipping.

Most likely that. Even though I disabled as much audio enhancement as I could find, something may still be enabled.

1 Like