I installed manjaro yesterday and since then I’ve been receiving lots of complaints that my mic quality is worse, especially on discord.
I used the default discord vc settings as a metric on both systems and even with voice suppression enabled on windows, the mic quality was still a lot better than manjaro.
Disabling noise suppression made my voice less “muffled” which is a start, however it still doesn’t explain the sudden loss in quality.
The loss in quality wasn’t limited to discord but it’s especially noticeable on discord.
I have a logitech g pro x wireless and I’m using the 3.5mm microphone that comes with it.
Launch Audacity, and record you microphone in the same conditions of volume (microphone volume to 100% in the system, no added BOOST) and of voice level, and microphone distance. Save as uncompressed format like WAVE or whatever. Do this the same on Windows and Manjaro. Compare these audio samples in an audio player.
Can you hear the difference?
Can you provide the samples for us to hear the difference?
I guess on Windows you have the whole G-HUB thing installed to provide drivers and settings/effects from Logitech?
//EDIT: maybe Solaar could help on Manjaro (not sure if it supports well your device)?
I have G-HUB installed on windows but I didn’t mess with my headphone/microphone settings, I only use it for my mouse. I also recorded the wav files in audacity but I couldn’t really hear much of a difference except for some more static in Manjaro and a slight loss in quality.
And sorry, I can’t upload a link to the wav files because the forum isn’t letting me for some reason.
I tested on steam and it was pretty close to audacity’s results. Also solaar supports some functionalities but it’s still pretty limited
I really think there’s actually a loss in quality but discord is probably worsening/amplifying this loss in quality.
From your two samples I would say, they are the same, they sound the same. Look at the files inside Audacity and compare the spectrum. You can see they have proper same level, the silence is clean no static here, quality sounds the same to me (I even though the Windows file was the Manjaro one initially).
I guess the problem lies in Discord and its configuration. Your friends are maybe trolling too, that’s a possibility knowing you switched to Linux.
You could do the test with someone you trust, and ask the person to record Discord output, and on your side you switch between Manjaro and Windows, so you can hear what it sounds like for the person at the end by listening yourself to the recording.
I already did that, I logged on a separate account on another device so I could hear my voice and there is a difference, primarily when krisp noise suppression is active.
My voice sounds muffled and certain short sentences or sentences that are a little naturally quiet will be cut out. (with krisp enabled)
Setting the noise suppression to “Standard” gets rid of most of the muffled stuff but it’s still kind of bad and disabling noise suppression completely makes it a little better but not by much.
If there was a way to capture my voice on discord I could probably show what I mean.
Since you’re on Plasma, open System Settings → Sound and see if the profile for the audio is being set to “Handsfree Head Unit” — Skype also does this, really annoying to be scolded for being in “Muffle Mode” before I have a chance to correct it!!!
From team meetings voicecall - the audio quality can differ a lot based on network latency.
Discord - as I recall, besides teamspeak - is the goto tool for in game voicecall - and so the network latency may - depending on load of course - be impacted - which then is blamed on Linux.
Ignore the slight cutout that happens at the end of every sentence in the windows ver as that doesn’t actually happen in VCs.
Apparently the windows ver. seems suppress the majority of the external noises (static, echoes, etc…) overall the voice is cleaner.
I already tried messing with the voice sensitivity in System Settings to see if it would pick up less static, echoes, etc but the quality is constant despite the sensitivity. The only difference is that discord wouldn’t pick my voice at all if I had some sort of noise suppression enabled and the sensitivity was too low.
I gave discord’s official client a shot and the Voice quality was pretty much the same compared to windows, my voice was also more responsive. Weird…
I also tested 3 different discord clients installed via flatpak and they all shared the same problems as vencord’s. The web based ones shared the same problems too…
Btw, does anyone know how to get the screen share audio to work? Apparently discord’s official client doesn’t support screen-share audio. Vencord managed to fix this but the streaming quality was pretty unstable and laggy.
I told you. Also, I didn’t know you were not using Discord, maybe making that clear from the beginning would have helped. Clearly, to me it was not clear… with sentences like that
VC like in Voice Chat?
and so on and so on…
I though the title was about Discord, and Vencord which I do not know about and which you did not talk about in your posts.
Since the beginning you say Discord, but you’re talking about another application, admit that it is not helping to understand.
Sorry for not making it more explicit, but I did include Vencord in the title to avoid this sort of confusion.
Anyways, from what I found this issue seems to be affecting only Arch based systems since there’s mostly just reports coming from those so there isn’t much that can be done here.
The default voice settings under User Settings > Voice & Video
Card sysdefault:3 'Heads'/'Logitech Logitech G PRO X 2 Gaming Heads at usb-0000:00:14.0-11.4.3, full speed'
Mixer name : 'USB Mixer'
Components : 'USB046d:0af7'
Controls : 6
Simple ctrls : 2
Simple mixer control 'PCM',0
Capabilities: pvolume pswitch pswitch-joined
Playback channels: Front Left - Front Right
Limits: Playback 0 - 74
Mono:
Front Left: Playback 63 [85%] [-11.00dB] [on]
Front Right: Playback 63 [85%] [-11.00dB] [on]
Simple mixer control 'Mic',0
Capabilities: cvolume cvolume-joined cswitch cswitch-joined
Capture channels: Mono
Limits: Capture 0 - 74
Mono: Capture 74 [100%] [0.00dB] [on]
The audio level meter in the videos show level is much lower in Windows, peak is approx 50%, but peaks in Manjaro are around 75%
Manjaro audio sounds very slightly distorted with faint background noise
To get more accurate values for levels, I downloaded the videos and extracted audio
using tlp-dl and loaded files in ocenaudio (an audio editor similar to audacity)
The True Peak levels from Manjaro (no noise su)-0.57 dBTP , -0.57 dBTP
this should be reduced by about 6-10 dB for optimal capture level
True Peak levels for Windows -19.53 dBTP , -19.53 dBTP
(this could possibly be increased by about 6-10dB)
I suggest open alsamixer to show microphone level control
alsamixer --card=`Heads` --view=capture
Reduce audio capture level by 9 dB and test microphone recording again
Level reduction should eliminate distortion and reduce background noise levels
You can HEAR audio levels are identical. He made tests without following a scientific procedure, himself being 1 inch further or closer to the mic in Windows or Manjaro will screw up scientific results you’re trying to investigate, same goes for the source, himself, if he speaks louder or quieter it will screw scientific results you’re trying to calculate.
He already found out there is no issue with his microphone, there is an issue with the NOT Discord various clients he tried. There is no microphone issue in the real Discord client.
Alsa Mixer though can still be good to open as indeed some hidden options could be found here. But again, “if it ain’t broke, don’t fix it”.
