A stereo compressor has one detector watching one signal. Whatever is loudest in the mix decides the gain reduction, and that gain reduction is applied to everything.
On most mixes the loudest thing is the kick, sometimes the snare, occasionally the lead vocal. So the kick sets the compression for the whole record. Every time it lands, the reverb tails duck, the cymbals dip, and the pads pull back. Do this gently and it’s the glue people like. Do it harder and the mix starts pumping in a way that has nothing to do with the music.
What splitting the detector does
Run the mid and the side through separate compressors with separate detectors and the kick only controls the mid.
The side channel — reverb, room, doubled guitars, stereo pads — gets compressed according to what’s happening in the side channel, which is a much steadier signal. It has no kick transient in it, so it doesn’t duck when the kick hits. The ambience keeps its shape while the centre gets controlled.
The audible result is a mix that feels tighter in the middle without feeling smaller around the edges. Most of what people describe as a compressor “squashing the life out of” a mix is the sides being ducked by centre transients, and this is the direct fix.
Settings that make sense per channel
The two halves want different treatment, because they contain different things.
Mid. This is where the transients are. Attack in the 10–30 ms range lets the kick and snare through before the compressor closes, which keeps the drums punchy. Release wants to be roughly in time with the track: fast enough to recover before the next beat, slow enough not to chatter. Ratio 2:1 to 4:1, and 3–5 dB of gain reduction is plenty on a mix bus.
Side. Almost no transients here, so a fast attack costs you nothing and a slower release keeps the ambience even. Ratio can be gentler, 1.5:1 to 2:1, with 1–3 dB of reduction. The goal on the side is consistency, not control.
Compressing the side harder than the mid is a specific effect and worth knowing: it brings the reverb tails up in the gaps, which makes a sparse mix feel fuller. It also makes the mix wider on average, since the side channel spends more time near its ceiling. Overdo it and the ambience starts breathing audibly.
Detector types matter more here
Because the two channels contain different material, the character of the detector matters more than it does on a single stereo bus.
An opto-style detector, with its program-dependent release, sits well on the side channel. It smooths ambience without the release ever calling attention to itself. A VCA or FET detector on the mid gives you the grip on transients that the centre needs. Mixing types across the two channels is one of the things mid/side compression makes possible and stereo compression doesn’t.
When to leave it in stereo
Mid/side compression is not an upgrade to be applied everywhere.
On a mix that’s already narrow — a solo piano, a mono-ish demo, a rap vocal over a centred beat — there’s very little in the side channel, and compressing near- silence just raises the noise floor and any stray room tone. You’ll hear it as a faint hiss coming up between phrases.
On material where the stereo image itself is the performance, like a live room recording or a well-placed pair of overheads, independent compression can loosen the relationship between the two channels and make the image drift. A linked stereo compressor keeps that relationship rigid, which is the correct behaviour there.
The tell is whether the two channels moving independently would sound like a mistake. On a drum room mic, it would. On a mix bus with reverb across it, it won’t.
Parallel is your safety net
Whatever you set, blend it back. A Mix control lets you compress the mid hard enough to control the centre, then dial it back until the transients return.
This matters more in mid/side than in stereo, because the two channels are being processed differently and heavy settings can start to pull the image around. Blending the processed signal against the dry one keeps the original stereo relationship underneath as an anchor. Set it hard, then find the point where the mix stops sounding processed.
A starting point
- Compress the mid first, with the side bypassed. Get the centre behaving.
- Bring the side in gently, aiming for 1–3 dB.
- Compare against bypass at matched level, since compression that’s just louder always sounds better.
- Check mono, because independent processing can shift the balance in ways that only show up when the sides cancel.