[pulseaudio-discuss] Mic Boost on AC97 audio causes bad behavior of mic volume control

[Wilck, Martin]

Hello,

there has been some discussion about Mic Boost  / Dock Mic Boost in the
past already. Some have reported no Mic output at all without Boost,
other say that enabling Boost will cause high noise on their systems.

I have another point here that AFAICS hasn't been reported yet.

It has been caused by the commit e6051cdf "alsa-mixer: Prefer moving
"Capture" before moving boosts"
(http://cgit.freedesktop.org/pulseaudio/pulseaudio/commit/?id=e6051cdf8db554c0bbd4257959c37a7ecc9c10c5)
(see also
https://bugs.launchpad.net/ubuntu/+source/pulseaudio/+bug/1085402).
 
In short, on my system, input volume control is hardly possible above
25% because of this patch.

The effect that I am seeing is as follows: I have two merged volume
controls for my Mic, "Capture" and "Dock Mic Boost". They are applied by
PA in the order encountered in the path conf file. On my system (Fujitsu
Lifebook E744 with Realtek ALC282 Codec), the "Capture" Amp has a range
of -18..30 dB in 0.75 dB steps, and "Dock Mic Boost" has 4 12dB steps
(0, 12, 24, 36). PA merges these two controls to a single volume control
with an overall range of -18..66dB. Normalizing at the max, PA converts
this to -84..0dB, where the Boost comes in at -36dB, or 25% of the
scale. The "base volume", calculated from the "offset" of the ALSA amp,
is at -66dB or 8%.

Because of the above patch, the "Capture" control is used first. Thus
the range from -84...-36 dB (or 0..25%, because of the non-linearity of
PA's volume control) is covered with "Capture". Above that level,
"Capture" is always held at 100% and only "Doc Mic Boost" changes. This
means that above 25%, there are only 3 steps that are actually possible:
-24dB / 40%, -12dB / 63%, and 0dB / 100%. 
=> There is no "smooth" volume control above 25%.
The graphical volume controls don't reflect this; thus user control
input volume is almost impossible at higher levels than 25%. Volume will
appear to change non-predictably to users.

By changing the ordering of "Capture" and "Mic Boost" in the path conf
file /usr/share/pulseaudio/alsa-mixer/paths/analog-input-dock-mic.conf,
I can get smooth changes over the whole dynamic range again. Therefore I
reckon that commit e6051cdf should be reverted because the effect
described here is general, while the noise level seems to be different
between models: On my system, there is no perceivable difference in
noise level if the overall volume is the same (thus, 30dB "Capture" +
0dB "Boost" has the same noise level as 18 dB "Capture" and 12 dB
"Boost" or -6dB "Capure" and 36dB "Boost").

However, I tend to think that PA's concept of volume "merging" generally
doesn't work well for a "Boost" control. The "Boost" is conceptually
rather a switch than a slider. A dynamic range of 84 dB doesn't make a
lot of sense for a cheap Mic like the typical laptop's Mic input. It
would be more reasonable to nail the Mic Boost switch to a certain value
that matches the hardware and just use the "Capture" dynamic range for
volume control. In the short run, this could be done by setting Mic
Boost to 0 or 1, e.g.:

[Element Dock Mic Boost]
required-any = any
switch = select
volume = 1
override-map.1 = all
override-map.2 = all-left,all-right

Ultimately, this shouldn't be hidden deeply in path conf files,
therefore it would be necessary in the future that PA exhibit the "Mic
Boost" control to users somewhere in the API / UI, unless some way is
found to "guess" the optimal value for the boost on different systems.

Please comment.

Regards
Martin