[Karl Uppiano]

"Walt" wrote in message
...
Eeyore wrote:
Radium wrote:
Karl Uppiano wrote:

Are you saying that
everyone has got it wrong for lo these many years?

Not necessarily. However, most stereo-to-mono conversion involve simply
decreasing the amplitude level by 50% and then downmixing to mono. The
problem with this, is that the stuff that was identical in both
channels is much louder than the stuff that was different in each
channel.

As was intended.

Sorry, but I've got to side with Radium here. The center channel build-up
when taking a mono sum is a real phenomenon, and *not* desirable or
intentional.

It happens because taking a voltage sum of two signals increases the level
by 6db, not 3 db as you might expect. For example, if the original
stereo recording has three singers at equal volume panned hard left, hard
right, and hard center, summing to mono will make the guy in the center 3
db louder than the other two. It was always thus.

So, what we have with Radium is a guy who likes mono (for whatever
reason - I'm not sure I want to know), but doesn't like how most stereo
programs sum to mono. So far, so good. Unfortunately his technique
doesn't come close to solving this problem - he gets .725(R) - .275(L) not
anything approaching a mono sum. But his problem is an understandable
one. (well, the sum-to-mono center channel buildup problem at least.
I'll refrain from commenting on the others)

What to do? Get used to listening in stereo? Write a signal processing
algorithm to compute a mono sum without the center channel buildup? (maybe
this has already been done?) Perform a mono sum the old fashioned way by
jamming a pencil eraser into one of your ears?

The possibilities are endless.

//Walt

I was thinking about this the other day, and it occurred to me that center
channel build-up is likely to be more of a problem with "fake" stereo --
multi solo tracks panned to their apparent position in the mix. A "real"
stereo performance, recorded live, with co-incident microphones probably
would not have this problem, although the performers at the center might be
louder due to their proximity to the microphone. That's probably one of the
reasons orchestras are often arranged in a semicircle.

I cannot think of a simple algebraic means to knock down the center channel,
without causing collateral damage to the un-correlated material in the left
and right channels.