You are not logged in.
I'm running Arch on a Dell Latitude 7300 (2019 model) since a few weeks, currently on kernel 5.5.5-arch1-1.
It uses a Realtek ALC3254 chipset for Audio playback. It seems to be correctly detected and playback works fine so far.
(This probably also applies to the Latitude 7400, which should have the same specs just with a bigger screen.)
However, when no sound is playing I can hear a little bit of white noise from the device speakers that I cannot explain. It's a quite high-pitched noise from the left speaker, while the right speaker is definitely a lower frequency white noise. It's only noticeable when it's quiet around or if you put your ears close to the speaker.
My machine is dual-boot and the noise is not there when I boot Windows 10.
I have tried a few things to figure out where this is coming from:
Installing pulseaudio did not change anything.
Putting options snd-hda-intel model=dell-headset-multi into /etc/modprobe.d/alsa-base.conf had no effect. (https://bbs.archlinux.org/viewtopic.php … 2#p1678732)
I have muted and unmuted outputs using alsamixer and have also turned off all captures on all devices. I also tried changing all gains to 0. The white noise did not disappear.
Interestingly, when muting the master output and/or the speaker output, the noise actually gets louder and goes back to more quiet, when un-muting. The gain seems to have no effect on the noise.
Manjaro, when run as live-system from USB behaves the same. However, when I run a live Ubuntu 19.10 from USB, the speakers work as intended and get properly muted (actually turned off?) a few seconds after the sound output stops, which also stops the white noise.
It seems like a pretty low-level driver/kernel/ALSA issue. Are there any other things I can try to get rid of the noise?
How can I get the Ubuntu-behavior of properly turning the speakers off when there is no sound output in Arch?
Thanks for your help.
Offline
I have a Dell Inspiron 15 7590 with the same audio chip and haven't noticed white noise, but I am having significant issues with the internal microphone and also microphone input via the headset jack. Have you had any issues with audio input?
Offline
So far I have only used the internal microphone with the volume adjusted via pulseaudio and that worked fine for Skype, Zoom and Jitsi conferences.
Last edited by dd23 (2020-03-27 17:09:15)
Offline
Could you tell me what version of the "linux" and "sof-firmware" packages you have?
BTW, my bug is being tracked here: https://github.com/thesofproject/linux/issues/1917
Offline
I have just updated to the following kernel:
Linux arch7300 5.5.13-arch1-1 #1 SMP PREEMPT Wed, 25 Mar 2020 16:04:40 +0000 x86_64 GNU/Linux
sof-firmware is currently not installed on my machine.
To see how things work, I have recorded a bit of audio with Audacity with both the internal mic and a headset connected via 4-pin 3,5mm jack and both work okay. However, if I adjust the levels such that you can properly hear my voice, there's noticeable static noise when I don't speak. It's really not ideal, but bearable, I'd say. The headset does a bit better than the internal mic, but still with noticeable static.
Edit: Cross-checked with Audacity on Windows (same machine, dual-boot). No noise whatsoever, neither on the internal mic nor on the headset. Must be a linux/firmware/ALSA issue.
Last edited by dd23 (2020-03-28 18:40:43)
Offline
OK, this sounds like a different issue to my machine. My internal mic isn't detected at all, because it requires some digital processing, which is only supported with the new SOF driver. Without the sof-firmware package installed, I get no audio at all.
You could try doing the opposite workaround to me, and *forcing* the kernel to use the SOF driver. There's a chance that might get it working for you. Basically, you'd need to follow this:
https://wiki.archlinux.org/index.php/De … t_detected
But I think you'd need dsp_driver=2.
Last edited by giddie (2020-03-30 11:09:28)
Offline
So, I played around with the kernel options and the SOF driver and here's what came out so far.
Installing sof-firmware and rebooting did nothing new. The same for setting options snd-intel-dspcfg dsp_driver to either 0 or 1. Nothing new meaning the same noise from the speakers and the same static when recording.
snd-intel-dspcfg dsp_driver=2 did not work at all, which makes sense, since this seems to be for Chromebooks, from what I can tell from here: https://github.com/torvalds/linux/blob/ … p-config.c
The interesting bit is snd-intel-dspcfg dsp_driver=3, which does load the SOF driver. With that I can select plenty of options for inputs and outputs in Audacity:
"speex" as recording device works for recording and does very well - no noise, good volume. Playback works without problems both with "upmix" and "vdownmix" as output device. Selecting "upmix: internal mic" for recording crashes Audacity. Minor annoying issue: swichting the input or output device takes a few seconds in which Audacity is completely frozen.
This also seems to completely bypass pulseaudio. I couldn't control any volume via pavucontrol. Running only SOF also fixed my initial problem of having quiet white noise on the speakers.
However, the default output seems to be set to something else since I couldn't get anything from Spotify, Firefox or Chrome when running SOF. I have not yet figured out how to select the default output in plain ALSA, which I assume would fix this problem.
I tried to control SOF via pulseaudio by running "pactl load-module module-detect" hoping that this would just work, but that just borught me back to my originial setup (white noise, noise while recording, similar to snd-intel-dspcfg dsp_driver=0 or 1). Killing PA with pulseaudio -k goes back to SOF.
If anyone has any suggestions, let me know. I'd be happy to test out other things or show some logs, debug info, etc.
Offline
Very interesting! You may need to go to the Pulseaudio project itself for support on something like this. I'm still stuck at the SOF driver stage, which is based at https://github.com/thesofproject/linux/issues, but it sounds like your driver might actually be loading OK now.
Offline
Just for the record: the SOF driver now works without problems for me.
In order to make ALSA use it as default (and thus also applications like my browser) I simply had to create /etc/asound.conf with the following content:
defaults.pcm.card 0
defaults.ctl.card 0
The parameter 0 was chosen because of:
cat /proc/asound/cards
0 [sofhdadsp ]: sof-hda-dsp - sof-hda-dsp
DellInc.-Latitude7300--0PYRY5
Maybe this helps people with similar issues.
Offline
It turns out the SOF bug was triggered only on systems with dual graphics cards. It's fixed in the latest sof 1.4.3 and 1.5.1. I've contacted the Arch package maintainer for sof-firmware, but the PKGBUILD needed a rewrite to get the latest versions working. It's working fine on my system now, though, with sof-firmware 1.4.3.
Offline