Ok I give up...anyone know how to extract center-panned stuff from a song?

I was messing with the Analog X vocal remover plug-in today. It's pretty cool. For those that don't know, it works by flipping the phase on one channel and then summing the L and R channel, thus canceling out anything that is panned dead center, and gradually getting to regular volume for stuff that is panned hard L or R. Anyway, I didn't like how it mixes the stereo down to mono, so I did some creative phase flipping and track summing in my DAW and figured out how to get the same center-canceling effect and preserve the stereo.

The other thing I wanted to try was getting the opposite effect, that is, keeping everything panned in the middle and eliminating anything that isn't common to both channels. However, after a good deal of messing with it, I just can't see any way to do this (in any analog or channel-summing method), without resorting to some kind of digital analysis of the signal which could then subtract out anything that wasn't common to both channels.

I might be staring the answer right in the nose, or be overlooking something very silly, but any thoughts would be appreciated.

One thing you can try: Do the cancellation of center data first, creating a stereo track (or 2 mono tracks, panned L&R - same thing in the end) without centered instrumentation; then, reverse the stereo track's phase & combine it with the original. That should cancel the instrumentation, leaving the centered info.

See, this doesn't quite work, because the non-centered track is actually the compliment of one side, plus the other side...meaning that if you take the non-centered track and add it to the flipped original stereo track, you will end up with double strength of the signal you flipped to get the non-centered track, and zero of the other side, which stayed untouvhed the whole time and is now cancelled by the original track flipped.

Am I right? Any other ideas?

With a surround plug-in who processes L-C-R, or with Dolby matrix hardware.

Example:

http://www.cubaselessons.co.uk/SurroundPlugs.htm

Surround Spreader

As usual, put in a 5.1 group. Send mono to center, or Stereo to L/R. Four Chorus units spread the input signal around the quadraphonic speakers.

Can be misused !

.

try using a M-S encoder plugin, then turning off the 'side' output.

Frankly, you are never going to get perfectly-isolated center tracks (it will be polluted with all stuff that is not panned hard right and hard left).

I've been wondering about the feasibility of a plugin that would allow extraction of arbitrarily-panned info from a stereo track - - For example, feed it a stereo mix, then only return stuff identified as panned to 10 o'clock (a bit left of center).

That would be a very useful mastering tool, methinks...

If you find anything like that out there, post agan & let us know, OK?

I'm not exactly sure what you're getting at, but I don't think you quite understood my response.

Let's say we have a recording of 3 sine waves: L is 400 Hz, R is 800 Hz, and in the center you have 1 kHz.

If you can remove the centered signal & create a new track, you are left with whatever is only L OR R side signal, i.e. 400 & 800. Then, if you take this and flip the phase 180 degrees & combine it with the original, the 400 & 800 Hz waves will be completely cancelled, leaving only the 1 kHz wave in the center.

L = 400 plus -400 = 0

C = 1000 plus -- =1000

R = 800 plus -800 = 0

The problem, however, is that most recorded signals, i.e. music, are not exclusively L,C,R - so you will have to make adjustments/compromises in practice.

franks analysis seems solid (both in terms of the process of grabbing compliment and the caveat of how well it'll work, but as you are currently using a "vocal zapper" you're prob well aware of this)

sadly, you are trying to "unsalt the soup" -- so, while useful, I'd expect results to be imperfect at best

Yes but you can't GET a 400 Hz and 800Hz signal from the original stereo track...since the way you cancel out the centered 1000Hz signal is by flipping one side and adding the other, your result is 400Hz + (-800Hz).

If you flip this 180 degrees then you have -400Hz + 800Hz.

If you add this to the original signal then you get

L = 400 plus -400 = 0

C = 1000 plus -- =1000

R = 800 plus +800 = 800 times 2 (+6dB)

ah, I see what your worried about - most of the HW zappr designs I've seen actually PRE EQ in the vocal range beforethe CMR (so as not to CMR bass stuff) -- does analog X not do this?

If so, how are they dealing with the bass freq?

Here's a theory that I think might work.

Analyze the source for the frequency content you're trying to keep. For instance for vocals we might try a range of 200 to 3K.

Create a copy of that stereo track with a notch filter to cut out information within that range. Now flip the phase of the copied tracks and sum them with the original. This will roughly cancel sound beyond those frquency ranges. Create a stereo mixdown from all 4 tracks.

Now apply the sum difference technique to this mixdown (1 sumed mono channel and a mono channel with the phase of one side flipped). Balance the volume of these 2 tracks to remove as much left over side data as possible.

While this will signiffigantly affect the sound quality of your intended source, it just might work.

OK, I see what you're saying. I didn't realize you were ending up with a track with one side out of phase after cancelling.

Perhaps it would be better to discuss using our 2 ACTUAL signals as opposed to a "virtual" center channel

so instead of L,R and C

we have R and L

(we can, perhaps, show the channels as a list of our comonents...so Common mode signals will be apparent AS will accidental comonalities)

you boys have too much time fiddling around with sinus waves

just subtract your mono, or center channel out of a stereo track, as we do it every day in surround production when only a stereo is available for a DD 5.1 *.ac3 stream:

http://www.york.ac.uk/inst/mustech/3d_audio/vst/welcome.html

B-Proc Ambisonic B-Format Processor

B-Zoom Ambisonic B-format Zoom control

B-Plane Mirror Ambisonic B-Format Processor

B-Mic Ambisonic B-format to Stereo Pair utility

.

http://www.dmalham.freeserve.co.uk/vst_ambisonics.html

B-pan Ambisonic Encoder

B-dec High resolution First Order Ambisonic B-format decoder

.

http://www.agmdigital.de/2/page6/page6.html

The ESsEX creates the most credible LCR from a Stereo source I have so far encountered.

.

leepmeister...

here one of the tools for you:

http://www.paulrharvey.co.uk/elevayta/product8.htm

.

Originally posted by Angelo Clematide you boys have too much time fiddling around with sinus waves .

Seem like a lot of sinus wave time spent there