2

What happens with the stereo image when compressing stereo program material using a linked dual compressor with a combined internal sidechain (a "stereo compressor")?

Is the stereo image reduced, unaffected, widened or what is going on?

EDIT: The question is not about wether stereo compressors are suitable for stereo signals - that is really the main premise for the design of such devices. The question is wether the stereo image is affected by the typical process of such devices (and if so, how).

Consider a hypothetical LCR mix:

  • Left: Drums
  • Center: Vocals
  • Right: Keyboard (strings)

The drums are punchy, the keyboard slow and "ambient" and the vocals of varying character tempowise.

These are then combined to a common mono sidechain which controls the compression of each channel. The stereo image should be intact, but is it really? Now you have something originally 100% left or right (keys and drums) interfering with the other channel too, through the sidechain signal. The transient reductions caused by the left side drums will be introduced in the right side keyboard (which originally had no activity at all related to drums).

  • 1
    I don't think so and I don't recall experiencing this, but I'm not sure and haven't found confirmation one way or the other online yet. I can't see how normal L/R stereo compression would ever widen an image. It most likely does not change it or narrows it. And it probably depends on the content. For example, a loud sound in the center that triggers the compression and brings down the overall level would reduce the level of quieter sounds that are panned wide. Stereo imaging can be changed by mid/side compression plugins or using M/S encoding and decoding around two unlinked compressors. – Todd Wilcox Nov 16 '15 at 15:08
  • 1
    This answer seems logical and indicates that the type of compressor and program material can cause any of the three results, but it's not authoritative enough for me to quote it in an answer: gearslutz.com/board/mastering-forum/… – Todd Wilcox Nov 16 '15 at 15:12
  • Good stuff in there - in particular answer #7: gearslutz.com/board/3706126-post7.html. I guess the answer depends on the definition of what makes something "stereo" - the difference alone, or the difference AND amplitude. The fact that the sidechain sources are rectified and summed to control two different signals tells me that the compression impact of the other channel somehow must influence the final stereo image.. – Michael Hansen Buur Nov 16 '15 at 15:24
  • Yeah. Now that I think about, it seems like any processing could affect the stereo image depending on the processing and the content. E.g., cutting lows with EQ could make an image sound wider, cutting highs maybe would sound narrower, etc. – Todd Wilcox Nov 16 '15 at 15:28
  • Yeah.. but with an EQ you do not have signal influenced processing bleed across channels over time, like with the compressor. You may be looking at a combined reference curve in some parametric interface, but the resulting filtering is static – Michael Hansen Buur Nov 16 '15 at 16:11
1

...Yes, as-stated, it would affect the stereo image, in your example, relative to without the sidechain (the panning positions won't change). The drum transients would be loudest for the most moments, creating a pumping effect through the mono sidechain, affecting the strings (and obviously the vocals, as they're centered), even though they're (100%?! really?) on the other channel. Therefore, the right channel will have some pumping, and the left will be causing most of it. Without the compressor's internal sidechain, unless you have a poor limiter on the Master track contributing to overcompression, there would be no pumping on either channel.

It would be your sidechain doing it, NOT inherently a compressor with the capability for linking both channels together.

But the real question is, why would you do it that way? It's not practical unless you were specifically emulating a live performance and you just didn't mic the drum kit with more than one mic. Furthermore, couldn't you just sequence a silent kick drum (sending with no volume to the Master) that lines up rhythmically with the original kick drum, and sidechains to the strings to make them duck?

And why not just try writing this yourself and seeing how it works in real life?


Below are some explanations on L/R, stereo, and M/S processing.

For instance, Density MKIII (a free VST plugin you can test) has the option of L/R processing, stereo processing, and M/S processing.

If you link the left and right channels during L/R processing, the knobs associated with L/R move in the same way at the same time as you adjust them. i.e. They are locked to be in "sync".

In contrast, stereo processing in that plugin allows you to affect both channels, but in differing ways if you so desire. You are not constrained to set one knob to X AND set its corresponding knob in the other channel also to X.

And just to be complete, M/S processing allows you to independently impose compression on the middle input (closer to fully mono in the stereo field), and the side input (closer to fully stereo in the stereo field).

A stereo-linked compressor therefore imposes a specific compression processing to both channel inputs in the exact same way, independent of what comes in. What comes out could sound different depending on what is in the stereo field, but that is usually going to happen anyway unless you have identical L/R channels.

  • Sorry - I'm pulling the accepted answer again - I updated the question with additional explanation. @audionuma you might want to adjust your answer too. – Michael Hansen Buur Nov 19 '15 at 12:46
  • added a few lines on my answer. Not sure if it will help. – audionuma Nov 20 '15 at 22:32
  • Thing is that if it is true for my extreme example of a mix, it wil also be true for less extreme mixes, with less apparent effect of course. I'm accepting your answer - but I'm not sure I understand this part: "It would be your sidechain doing it, NOT inherently your stereo-linked compressor." - If not the combined (internal) sidechain, what is it then about stereo linked compressors that makes them stereo linked? – Michael Hansen Buur Nov 24 '15 at 10:05
  • Yes, the internal sidechain feeds the signal of the drums into the strings, but a compressor with the stereo-linking capability, if designed well, would have the option to turn the sidechain off. Maybe I can clear up my wording by saying "compressor with the capability for stereo-linking". (Also, you did say, "a linked dual compressor with a combined internal sidechain," which seems to imply both qualities on one compressor, each of which are supposedly different.) – timaeus222 Nov 25 '15 at 7:39
3

As you specify using a 'stereo linked compressor', neither the timing relation or the level differences between both channels should be modified. Therefore, the stereo width should be unaffected. That's what 'stereo linked compressors' are made for.

Nevertheless, there could be unexpected perceptive differences, especially if you're hitting hard into the compressor and modifying transients for example. The reproduction system (headphones, TV loudspeakers, Hi end monitoring) could make these artifacts more or less perceptible.

[edit] Taking the (extreme) example of LCR mix given by the OP, and assuming that having a drumkit full left and a keyboard full right is a stereo image (which could be discussed) :

before processing

drums : full left

vocals : center

keyboards : full right

after processing (given that same gain reduction is applied to all channels)

drums : full left

vocals : center

keyboards : full right

[/edit]

  • Left channel: drums. Right channel synth. These two are summed to mono and used to compress both channels... It seems counterintuitive that the stereo image will not change since the drums from the sum will pump through on the strings. However the processing of the sidechain will definitely mean something here.. – Michael Hansen Buur Nov 17 '15 at 7:09
  • Well I probably misunderstood your question as to me the input to the processor is a stereo signal, not a double mono one. Can delete my answer if that's the case. – audionuma Nov 17 '15 at 7:30
  • No.. that's not what I meant. I meant a stereo signal. The example above is just a rather extreme case 😆 – Michael Hansen Buur Nov 17 '15 at 9:35
  • Doesn't stereo implies a sort of correlation between the left and right channel ? I don't see the definition of stereo image (before or after processing) in the extreme case you mention. – audionuma Nov 17 '15 at 12:55
  • 2
    @michael There will be no effect on the stereo image. Both sides of the synth would be effected the same, it won't change their left/right panning st all. That's not to say it won't effect the sound. Of course the synth would pump with the drums - I'm hoping you're just using that as a hypothetical:) - but both sides will pump the same. If you record, say, stereo piano and forget to link then then the sides would be compressed separately causing your stereo piano to lean back and forth when the compressors are not in sync (i.e. one is compressing more than the other). – Tom Nov 17 '15 at 15:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.