What is mid/side processing?

Mid/side (M/S) processing is a way of treating a stereo signal as a centre channel (mid) and an edges channel (side) instead of as left and right. Per Wikipedia's article on joint encoding, the mid channel is the sum of left and right, the side channel is their difference, and, per Wikipedia's article on stereophonic sound, left and right come back by adding or subtracting the two.

So anything identical in both speakers lives in the mid. Anything that differs between them, such as wide reverb, hard-panned doubles or room ambience, lives in the side.

How it works

1. Encode. Mid = (left + right) / 2; side = (left minus right) / 2.
2. Process. Put an EQ, compressor or plain gain on one channel only.
3. Decode. Left = mid + side; right = mid minus side (Stereophonic sound, Wikipedia).

With no processing in between, the round trip returns the original left and right exactly; the halving in the encode is what keeps the gain unchanged. Sources differ on where that factor of 2 goes: drop it from both steps and you get each channel back at double level (+6 dB), which is still lossless up to a fixed gain factor. That is why codecs use it: Wikipedia notes MP3 encoders in joint stereo mode switch between M/S and L/R frame by frame, and lossless codecs such as FLAC use M/S because it introduces no artefacts (Joint encoding).

Where it came from

M/S started as a microphone technique: a sideways-facing figure-of-eight mic for the side, plus a forward-facing mic, usually a cardioid, for the mid. Wikipedia says this produces a completely mono-compatible signal, and that recording mid and side separately lets you change stereo width afterwards by adjusting the level of the side signal (Stereophonic sound). FM stereo broadcasting uses the same idea, sending the sum so mono receivers still work (Joint encoding).

Common uses in a mix or master

• Width: raise or lower side gain. In our experience, many "stereo width" controls work this way.
• Low-end focus: a high-pass filter on the side keeps bass in the centre.
• Vocal space: a small cut on the mid where the vocal sits, instead of on the whole bus.
• Taming harsh edges: de-essing or compressing only the side when cymbals or reverbs spray.

For panning and width in general, see stereo width and panning; for checking the result, see mixing in mono.

What we think

M/S is most useful as a diagnostic before it is a tool. Solo the side channel once on any mix: you will hear exactly what disappears in mono, and usually find bass or a lead sitting there that should not be. Our view is that small, corrective moves (filter the side lows, nudge side level a decibel or two) beat dramatic widening, because the side is precisely the part that vanishes when a phone speaker or club system sums to mono. If a track only sounds big with the side pushed hard, the arrangement probably needs more contrast instead. That is our judgement, not a measured result.