12 Nov
2010
12 Nov
'10
10:18 a.m.
On Thu, Nov 11, 2010 at 08:22:08PM -0800, Eric Kampman wrote:
I looked up "pan law" and understand that
center should be
-3 db (or some say -3.5 or 4.5, whatever) given unity at
panned hard L or R. It was said that in an ideal room I
should be down -6 db in the center. That would mean linearly
transitioning from unity gain to completely off as one pans,
I *think* (-6 db in the center = .5).
The ideal value depends on frequency, in theory -6 dB of LF
and -3 dB for HF, with a gradual transition at around 500 Hz
or so.
I played around this for awhile and using the sum of
the square of sin and cos etc I got an f(t) of
f(t) = cos( pi * t / 2 ) (details available on request)
So L(t) = cos(t * pi / 2) and R(t) = cos((1 - t) * pi / 2)
And that seems to work out correctly.
It is correct.
Is this equal power version worth spending the
processing cycles on?
I intend to make pan envelope and LFO controllable so it's not going
to be the case that the pan value can be thought of as relatively static.
If you are concerned about the CPU use, consider this:
panning value x = [-1..+1]
L_gain = (1 - x) * (0.7 + 0.2 * x)
R_gain = (1 + x) * (0.7 - 0.2 * x)
which will be -3.1 dB at the center. You can modify this
modyfying the 0.7 and 0.2 constants, for example
L_gain = (1 - x) * (0.65 + 0.15 * x)
R_gain = (1 + x) * (0.65 - 0.15 * x)
will produce -3.75 dB, and
L_gain = (1 - x) * (0.6 + 0.1 * x)
R_gain = (1 + x) * (0.6 - 0.1 * x)
will produce -4.4 dB, etc.
Ciao,
--
FA
There are three of them, and Alleline.