How to lower the volume of the center of a mix to free some room for vocals ?

He may have used a Mid/Side splitter, separating the components that occupy the center from the those that live out on the sides and dropping the levels on the Mid portion. As suggested above, Channel Tools is the tool for that. You could also grab Voxengo's free MSED plugin, which is a little simpler to use.

Another option isn't to lower the center, but raise the sides to create more "space" for whatever is in the center. You can do this with image widening and binaural panning plug-ins (e.g., the Blue Tubes stereo imager). This is effective if the space you need to open up isn't too huge.

cparmerlee
My only confusion about these plug-ins that support mid-side is that they don't really give you much (or any) control over where the boundary between mid and side lies.  I think Channel Tools can do that, so I guess you could create two buses and use Channel Tools to separate mid from side, then apply EQ to the mid bus, which I think is what Bitflipper is suggesting.

That's because there is no boundary as such. "Mid" means all the components that are the same on both the left and the right channels. "Side" means all the components that are different between left and right. Subtract left from right and multiply by 2 and what you're left with is whatever they had in common. It's a straight-up arithmetic operation, not a gradual transition.
 
Of course, most components are not exclusively left, right or center. A mono track panned 50% left still has content in the right channel, so is it going to show up in the Mid or the Side? This, I think, is the confusion you're referring to.
 
The answer is that it's both Mid and Side. Whatever portion of that track's amplitude that is common to both channels falls into the Mid component and the part of it that's different falls into the Side component. Remember, we're not talking about discrete instruments here, but percentages of amplitude across the frequency spectrum. In this example, you'll still hear that track when you solo either the Mid or Side, because it's both.
 
So what happens to that track when you raise the Side component using Channel Tools or MSED? Assuming the left and right channels are spectrally identical, you'll simply be exaggerating the L/R volume difference. In other words, panning it further to the left. If it had been a one-track mix, you could have achieved the same end with the pan slider.
 
But this track is likely part of a whole mix consisting of other elements that are each panned differently and that have different spectral content. Consequently, raising the Side is more complex than simply exaggerating all the pan positions in your mix. It's also going to redistribute frequencies across the panorama, because some frequencies are going to be more similar left-to-right than others. Effects such as reverb (both artificial and natural), chorus and delays all contribute to subtle left/right imbalances that will be emphasized by raising the Side.
 
On a full mix, making room for a vocal by lowering the Mid portion often makes more sense than raising the Side. But using M/S EQ makes even more sense. Why turn down center-panned frequencies that don't conflict with the vocal? You'd just be losing things like kick and bass that weren't hurting anyone to begin with, and are necessary to keep the energy up. With Mid/Side EQ, you can lower just those frequencies that would otherwise mask the vocal, while keeping the lows and highs intact.
 
If you want to take it one step further, there are plugins that transparently make room for the vocal by dynamically dipping only conflicting frequencies and leaving everything else alone. 
 
 
 
 

sharke
I would think that if you're going to do this, it's probably a good idea to make sure that your "sides" are very mono compatible, because if they're not (i.e. they sound phasey when collapsed to mono) and you're attenuating the solid "mid" component then I suspect it would make things sound even less solid in mono.

One great aspect of binaural panners is they collapse properly into mono, because of the way the accentuation is derived in the first place from a mono signal.

Another way to make room for vocals is to use dynamic EQ. So when a threshold level in a certain spectral range is met then the equaliser will dip or expand. Using Mid/Side assignments you can decide if you want to increase the mono component of the vocal track or decrease the sides in that area. Note, this only triggers when the vocal happens.
 
Also it's worth noting you can do this with Cakewalk LP EQ however I don't feel my aging PC can handle the power it requires especially if there is lots of other things going on.
 

The first one I ever saw was Elevayta's Space Boy, a little dated now - AFAIK it was never ported to 64-bit - but still available, and cheap.
 
What I use for this is Meldaproduction's MSpectralDynamics. That's not the main thing MSpectralDynamics does, just one of its more obscure talents. It's also the most transparent one that I've tried.
 
All these plugins work the same way: a sidechain signal is analyzed and its spectrum is inverted for the main signal's EQ. You set it up the same way you'd set up a sidechained compressor.
 
A less-expensive alternative would be MAutoDynamicEQ, which I see is currently on sale for 50 EUR. It's a superb dynamic equalizer (also an excellent all-around EQ) that supports a sidechain input, and it's also adept at EQ-matching and (more helpful) EQ inversion, e.g. analyzing a signal and inverting its spectrum to apply to another signal.
 
Another one is Fabfilter's Pro-MB. It's a cross between a dynamic EQ and a multiband compressor. For carving EQ, you want to be sure it's in Minimum Phase mode.