[pulseaudio-tickets] [Bug 97777] New: Clipping with per-stream volume above 100% even though device volume below 100%

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=97777

            Bug ID: 97777
           Summary: Clipping with per-stream volume above 100% even though
                    device volume below 100%
           Product: PulseAudio
           Version: unspecified
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: normal
          Priority: medium
         Component: daemon
          Assignee: pulseaudio-bugs at lists.freedesktop.org
          Reporter: bugs.freedesktop at haasn.xyz
        QA Contact: pulseaudio-bugs at lists.freedesktop.org
                CC: lennart at poettering.net

I am doing the following:

1. Set my device volume to about 70% (approx. -10 dB gain)
2. Play a sine wave at -0 dBFS using a simple test client:

λ import Sound.Pulse.Simple 
λ p <- simpleNew Nothing "pulse-simple" Play Nothing "testing" (SampleSpec (F32
LittleEndian) 48000 2) Nothing Nothing
λ simpleWrite p [ 1.0 * sin (w*x) | x <- [0..48000*5], let w = 0.01 :: Float ]
λ ...
λ simpleFree p

3. Change the per-stream volume for this stream (“pulse-simple” in pavucontrol)

Any value below 100% is fine, but as soon as I go above 100% even by a little
bit (e.g. 101%), I immediately hear obvious clipping artifacts.

4. Set the per-stream volume to 100%, and change the amplitude of the sine wave
produced instead (e.g. changing 1.0 to 1.1). I again hear obvious clipping
artifacts (even though the signal format is floating point).

Why is this the case? Shouldn't the overall volume of 70% give me more than
enough headroom to go above 100% on the per-stream volume? The only explanation
I can come up with is that the conversion to integer happens before the device
volume adjustment, e.g. perhaps if the device volume is being applied by ALSA
instead of by PulseAudio (since it does seem that changes to the device volume
are also visible in alsamixer).

If this is indeed the case and this is a limitation of the way device volumes
are implemented, would it be possible to apply some sort of global “gain”
setting to all sinks instead? I want to give them all 10 dB or so of mixing
headroom so I can use software volume controls to go above 100% without
clipping.

Obviously I can achieve this by setting up some sort of LADPSA filter or
whatever in the signal path of every single stream, but that seems like a
hassle to set up properly (especially since I'm using multiple audio devices).
If it would be possible to just add a simple “mixing headroom” variable to
daemon.conf that I can set to get headroom for free, I would be a very happy
man.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20160912/2aebc6c5/attachment.html>