Ducking a piano M/S signal when the vocal is on?

Hi -

I have a stereo piano track and a vocal. What I’d like to do is to duck the mid signal of the piano (talking M/S kind of mid here) whenever the vocal is happening (the vocal is front and center), just as one more tool to increase vocal clarity. I know one way to do it is to have the vocal feed the side chain of a comp that is set to compress the M signal, and adjust it so there is 1 or 2 dB of gain reduction in the piano M signal whenever the vocal is feeding the side chain. I was reading that a drawback of that is that it may work the least when it is needed most, and vice versa. For example, when the voice is softest, we may want the most ducking of the piano M signal, but this is the time the gain reduction is actually the least.

Because of that, and because I don’t own a comp that does M/S and have a side chain, I tried another method:

I sent the piano to a group with the group’s first insert slot being a plug in that filtered out the side signal so only the mid was left (I used SPAN). In the 2nd slot I put a gate, and fed the gate’s side chain with the vocal. So, every time the vocal occurred, the gate opened and some of the piano M signal was passed; whenever there was no vocal, the gate was closed, and nothing was passed.

I then routed that to another group for the sole purpose of inverting phase of the piano’s mid signal, and then on to the master bus. So vocal caused cancellation of the piano’s mid signal, to a degree controlled by …

Well, it worked well in the end, but I hard a very hard time precisely controlling the amount of dB reduction in the M piano signal using any combination of the gate’s threshold and/or the send level of the vocal to the gate’s side chain. It was either a huge drop in the piano M signal volume, or no drop at all in my hands. In the end I found the best way to control that precisely/delicately was to adjust the post-gate gain feeding the phase reversal (I used the gain control in the pre-section of the “Phase Reverse Group” immediately after the gate). It wasn’t hard at all after that to dial in a 1 or 2 dB drop in the piano mid signal every time the vocal came in. It was fun to “check the ears” by putting another instance of SPAN on the master bus to show the piano M signal only, pulling the fader down on the vocal, and watching the piano M drop a dB or two every time the gate opened.

Anyone have any thoughts or suggestions on this? It occurred to me it would have been a whole lot easier just to automate the amount of mid compression on my Fairchild 670 so there was a dB or two of compression of the piano’s MID signal every time the vocal came in, but of course that wouldn’t get around the possible problem of having the least M signal piano gain reduction when the voice was the softest (i.e., when probably I’d want the most gain reduction).

(Also, just to confirm … the Cubase gate is sample accurate - no phasing when the parallel signal was run through a wide open gate, and complete nulling with phase reversal).

Thanks for any thoughts!

Compressing the signal that is feeding the side chain will add more stability and predictability. Doesn’t mean that you will not need to ride faders though. :slight_smile:

Thank you for that, Tom!

I’ll think hard about that, I have to admit it’s hard for me to work through these things! I guess compressing the side chain would make it so it would be more likely that all vocals would open the gate, sort of a toggle effect? Right now I have the threshold set so low that essentially any vocal opens the gate. I wonder if that’s the same end result …?

To be honest, I’d love it if there were a way to have it set up so the piano got ducked only during soft vocal passages, and not so much during the louder ones. Haven’t figured out how that can get done, though … oh, that’s what a fader is for! :smiley:

Thanks again for your comment!

Trial and error is your friend here. Also, some compressors act differently than others; if you aren’t getting the results you seek with comp A, try comp B.

And I agree completely that riding faders is a good way to go. In fact, most really good engineers will choose to ride faders first, and use a comp as plan B.


Thanks, Jeff. Well, it was a great exercise to set that up, and maybe one day I’ll find a situation where it is the best answer to some problem. But having done it, it does seem somewhat over complicated, especially since the amount of phase inversion will need to be adjusted with fader rides anyway.

So, I have just built a much simpler way … put MSED (in “INLINE” mode) on the source audio, and fader ride the Mid Gain as needed. No multiple groups, gates, thresholds, etc. I guess the automated way (comp, gate, etc.) is for people who have 14 projects to get out the door by noon … quick and easy (once it’s set up), and gets you 90% of the way there. I don’t have any of those time constraints, just a hobbyist who works on one project at a time (if I remain self-disciplined!).

Thanks again, everybody -