Front microphone input boost on NUC8i7BEH under Linux

[pnc]

Hi! I have a weird question and could some additional data to help troubleshoot or fix an issue I have with my NUC (NUC8i7BEH, but keep reading even if you are on another NUC.)

tl;dr if you scroll, I just need the output from a few hardware info commands if you use a NUC and Linux.

I use Debian 10.5 with a very default PulseAudio audio. The built-in 3.5mm jack on the front works shockingly well (it detects the presence of a plug in the jack) and I use it with a TRRS (combo mic/headphone jack) headset for conferencing and stuff, and input and output work OK.

However, the audio chipset on this NUC (the Realtek ALC233) exposes both a microphone input gain level as well as a microphone "boost" level. ALSA sees both of these and labels them as "Mic" and "Mic Boost." They work roughly like this:

Mic: normal slider, -34dB to +12dB
Mic Boost: notched slider with 0, 10, 20, and 30dB boost levels (supposedly.)

However, the Mic Boost only seems to work at the 0 and 10 settings. If you change it to 20 with alsamixer, there is a pop (if you're recording) and then… no sound at all. Ditto 30dB. Just nothing at all, no microphone input. This actually matches most of the "Mic Boost" settings I've seen on soundcards, which are just a checkbox—no boost, or +20dB boost.

Anyway, this would be fine except PulseAudio "merges" the Mic Boost onto the Mic volume slider, and so at a system level (Zoom, Chrome, Gnome settings), there's a single mic volume, which works perfectly from 0-50%, gets louder at 50-62.5%, and then makes a large pop at 62.6% and is completely silent from there up to 100%. (Confused? Imagine how long it took me to figure out this was happening as Zoom and Chrome automatically adjusted the input volume for auto-gain.)

I made a configuration tweak to PulseAudio to make it not merge the two sliders, and I can set the boost manually now, and everything works great.

Anyway, evidently the usual thing to do for weird hardware bugs like this is to submit a patch to ALSA's Realtek codec, which contains about one billion other little workarounds for goofy hardware.

Before I do this, I want to confirm I'm not the only person with this problem, and that I don't just have bum hardware.

tl;dr: scroll here!

Bored Linux NUC users, could you reply with the output of the following commands to help me dig into an audio issue?

cat /etc/issue # Distribution info if present
uname -a # Kernel version
pactl list sources # List what PulseAudio knows about
cat /proc/asound/card?/codec#? # ALSA dump for all cards, all codecs

If you have an extra 5 minutes to spare and a 3.5mm jack headset, I'd be curious to know if you see the same behavior around the "Mic Boost" setting I describe above, or if I'm alone here. (Even if you're on a different model NUC, I'm actually even more curious—if many people are affected, it's even more worth patching in the kernel rather than only working around it in PulseAudio.) You can reproduce it by plugging in a headset (old Apple headphones work fine), opening up any sound recording app, and then adjusting the input volume wherever you normally do that while speaking, to see if the input cuts out at any point.

Windows NUC users with 5 minutes to spare and a 3.5mm jack headset and are using the official Realtek drivers, do you have a "Mic Boost" setting anywhere in the Windows sound mixer? How many levels does it have? Do all the levels work with the front input jack? Does the input drop out briefly when you change them? A screenshot of the mixer would be killer.

Thanks a ton!