[ardour-dev] playback levels

[Florian Schmidt]

On Wed, 25 Feb 2004 10:17:49 -0600
Bill Gribble <grib at billgribble.com> wrote:

> In my experience the purpose of the panning gain curve is to provide
> for perceptual "sameness" of volume throughout the panning range when
> putting a mono input on a multichannel bus.  Visualize a full-range
> mono signal panned left, then center, then right.  With no gain
> adjustment, the total signal has twice the amplitude when panned
> center as it does when panned hard right or left (that's a 6dB boost).
>  That's just wrong

Hmm, let me see, if i understand that right. Let's assume a mono file on
a mono track that gets played back to a stereo output. Let's also assume
that this mono file contains a 0db signal. 

Now if i pan this 0db mono track to center i would think that the
desirable behaviour is to actually "copy" the mono file at full volume
to both stereo output channels.. So both output channels now carry a 0db
signal..

Now, when i pan this track all the way to the left, i would want to have
the mono signal on my left channel but boosted by +6db and i would like
to have silence on the right channel. If this busts my headroom, i need
to adjust manually.. My point really was that i don't want having
different volume levels when having panning enabled as opposed to having
it disabled.. I'm with you that there needs to be more done than just
cuttting one of the output channels for panning.,,

Now the case with a stereo track is a little more complicated
[especially as i start to think about it in more detail now].. A stereo
track with a 0db signal on each of its channels should just be copied
1:1 to the output stereo channels when panned to the center.. But what
happens, when it is panned all the way to the left? Several scenarios
are thinkable:

1] The left channel is boosted by 6db and the right channel is just
silenced. In this case the "panning" is done by just lowering one
channel whhile boosting the other.. The "drawback" is: You only hear one
channel of a stereo source when panned totally to one side.. This is
basically the same as splitting the stereo file into two mono files, one
panned fully to the right and one panned fully to the left. COntrolling
the "placement" in the mix is then done by just playing with the
individual channel volumes and the panner just provides a convenient
interface to do this..

2] Both source channels are added and sent to the left output channel.
the right channel is again silent.. Interesting is the transition from
center to full pan on one side.. At 1/3 pan to the left, 1/3 of the
right channel signal is added to the left channel and 1/3 of the right
channel is also substracted from the right channel [leaving the right
channel with 2/3 of its original volume]..  The goal of this method is
to have both channels of the source stereo file to be always audible in
the mix while still preserving some of the stereo image of the stereo
source in the not_fully_panned cases..

3] The stereo source is always mixed together into a mono source and
that is then positioned in the mix.. This case, of course, throws away
all stereo info from the source and just positions the added signal
somewhere in the stereo mix.

I know too little about perception of volumes in stereo signals to
really judge it, but i suppose that a stereo track with 0 db signal in
both channels is probably about as loud as both channels mixed [added]
together and only sent to one output.. Though i can imagine that this
differs for different kind of signals. For example a same-phase-sinus
signal in both channels will produce different percepted volume changes
when panning than a stereo signal of, for example, a piano on the left
channel and a bass on the right.. 

Maybe i also talk out of my ass with the above stuff ;) I never really
gave this much thought up to know. I still don't see though, why a
panned signal should be lowered by 6db per default..

-- 
signature :)