M/S processing explained

the hidden dimension

your mix lives in left and right. every fader, every pan knob, every bus operates in the L/R domain. but there is another way to see the same signal: centre and sides. think of it like photography layers: mid is the subject, side is the background. M/S processing splits your stereo field into what both speakers share (Mid) and what makes them different (Side), and lets you process each independently.

this is not a new idea. Alan Blumlein described the principle in a 1933 British patent, originally for cutting stereo records with less groove wear.[^1] Holger Lauridsen built portable M/S matrix devices for Danish State Radio in the 1950s, because M/S encoding guaranteed mono compatibility for listeners with single-speaker radios. the technique’s usefulness has only grown since then.

what Mid and Side actually contain

understanding what lives in each channel is the key to using M/S processing effectively.

Mid contains everything that is identical in both speakers. in a typical mix, that means: the lead vocal (usually center-panned), the bass, the kick drum, the snare body, and any other element sitting at the center of the stereo image. if you solo the Mid channel, it sounds like a mono fold-down of your mix, because that is exactly what it is.

Side contains everything that differs between the left and right channels. that means: reverb tails, stereo-recorded instruments (drum overheads, room mics), hard-panned elements, synth pads with stereo spread, and any processing that has created differences between L and R. if you solo the Side channel, it sounds thin and ghostly, because center elements have been removed.

a purely mono signal (L = R) produces zero Side content. a signal panned hard left produces equal Mid and Side: L = signal, R = 0, so Mid = signal/2 and Side = signal/2.

why M/S does things L/R cannot

the power of M/S processing is access to a dimension that left/right processing cannot reach. with L/R processing, turning down the right channel removes everything on the right: the right side of your reverb, the right-panned guitar, and the right-channel portion of your center vocal. there is no way to target “the center” or “the sides” independently.

with M/S, you can:

widen the reverb without making the vocal louder. boost the Side channel above 2 kHz. the reverb tails and stereo ambience get more prominent. the center-panned vocal, which lives entirely in Mid, is untouched.

narrow the image without losing the center. reduce the Side level. the stereo spread pulls inward while the vocal, bass, and kick remain at full strength.

EQ the center vocal without affecting the stereo guitars. cut 200 to 400 Hz on the Mid channel only. the muddiness in the center clears up. the stereo guitars, which live partly in Side, keep their body.

mono the bass without touching the rest of the stereo field. high-pass the Side channel at 100 to 150 Hz. everything below that frequency collapses to mono. everything above retains its full stereo width.

five M/S techniques

side EQ

the most common M/S technique. high-pass the Side channel at 100 to 200 Hz to mono the bass. boost the Side at 8 to 12 kHz for stereo “air” and sparkle. cut the Side at 2 to 4 kHz to reduce harsh stereo content like cymbals or bright guitars. every mastering engineer uses Side EQ.

side compression

compress the Side channel to control stereo image dynamics. this prevents the width from swinging too much between quiet and loud sections. useful in mastering when the stereo image is inconsistent: verses feel narrow, choruses feel too wide. a gentle 2:1 ratio on the Side channel evens it out.

mid-side balance

the simplest form of M/S processing: change the relative level of Mid and Side. reduce Side relative to Mid and the image narrows. increase Side and it widens. at unity, the original balance is preserved. going beyond unity “hyper-widens” the image, which works in small doses but pushes the correlation toward zero.

mid EQ

EQ the center-panned elements without affecting the stereo field. cut muddiness at 300 Hz in the center without dulling the stereo guitars. boost presence at 3 kHz on the vocal without brightening the cymbals. this is the inverse of Side EQ and equally useful.

multiband M/S

different M/S balance per frequency band. mono below 200 Hz, original balance from 200 Hz to 5 kHz, wider above 5 kHz. this gives you frequency-dependent width control, which is more precise than a single Side level change.

the gotchas

boosting Side does not create width. it amplifies existing differences between left and right. if there is no Side content (a mono source), boosting Side does nothing. M/S processing cannot create width from mono. it can only manipulate existing width. to create width from a mono source, you need decorrelation or delay.

excessive Side boost causes correlation problems. pushing Side above unity tips the correlation toward zero or negative. below +0.3 is the caution zone. below 0 means mono playback will cause audible cancellation. check your correlation meter.

M/S is not the same as left/right. a common confusion. Mid is not “center panning.” it is everything common to both channels. a hard-panned element appears in both Mid and Side equally. do not think of M/S as a different kind of panning.

phase-linear EQ matters more in M/S. if your Mid and Side EQ introduce different phase shifts, the recombined L/R will have subtle spatial artifacts. linear-phase EQ avoids this but adds latency and pre-ringing. for most mixing applications, minimum-phase EQ is fine. for mastering, consider the trade-off.

subtle moves only on the master bus. 1 to 2 dB of M/S EQ sounds professional. large moves sound broken. if you need more than about 3 dB of M/S adjustment on the master, the problem is better fixed in the mix.

checking your work

M/S processing gives you two powerful diagnostic tools built into the technique itself.

solo the Mid channel. this is your mono fold-down. if the Mid solo sounds bad, thin, unbalanced, or missing critical elements, your mix has mono compatibility issues. what you hear in the Mid channel is exactly what will play on every mono speaker.

solo the Side channel. this is everything that cancels in mono. if you hear critical elements in the Side solo (a vocal, a bass note, a snare), those elements will lose energy on mono playback. the Side solo reveals phase problems, unintended stereo artifacts, and elements that are wider than you realized.

use a correlation meter to monitor continuously. a full mix should sit between +0.3 and +1.0. individual tracks can dip lower, but the master bus should stay safely positive. if your M/S processing pushes the correlation below +0.3, back off the Side level or reduce the widening.

frequently asked questions

frequently asked questions

what is mid/side processing in audio?

mid/side processing is a technique that splits a stereo signal into two components: Mid (everything the left and right channels share) and Side (everything that differs between them). you can process each independently, then recombine them back to left/right. it lets you EQ, compress, or adjust the center image and stereo field separately, which normal left/right processing cannot do.

how do you calculate mid and side from left and right?

the encoding is simple: Mid = (L + R) / 2, Side = (L - R) / 2. to decode back: L = Mid + Side, R = Mid - Side. the round-trip is perfectly lossless. no information is created or destroyed. the division by 2 preserves the original gain level.

can M/S processing create stereo width from a mono source?

no. M/S processing can only manipulate existing stereo content. a mono signal has identical left and right channels, which means the Side channel is zero. boosting the Side channel of a mono signal does nothing, because you are amplifying silence. to create width from a mono source, you need a decorrelation technique like allpass filtering or Haas delay.

how much M/S EQ is too much on the master bus?

on the master bus, keep M/S EQ moves to 1 to 2 dB. if you need more than about 3 dB of M/S EQ to fix a problem, the issue is better addressed in the mix itself. large M/S moves cause spatial artifacts, shift the stereo image in unnatural ways, and can push the correlation toward dangerous territory.

what is the difference between M/S processing and stereo widening?

M/S processing is a tool that lets you independently adjust the center and sides of an existing stereo signal. it is one way to control width (by changing the Side level), but it is not a stereo widening technique on its own. true stereo widening creates new differences between left and right, using techniques like decorrelation or delay. M/S only redistributes what is already there.

references

a note from the developer

M/S processing is one of those techniques that sounds intimidating until you try it. the math is simple: add the channels, subtract the channels, process independently, recombine. the results can be dramatic, especially the first time you solo the Side channel and hear everything in your mix that differs between left and right.

when i built the M/S mode in KERN WIDE, the most important decision was what not to do. M/S mode does not apply the bass crossover that STEREO and HAAS modes use. the reason: M/S mode boosts existing Side content. if you first collapse the bass to mono (removing the Side content below the crossover), then try to boost the Side, you are amplifying the signal you just removed. it is a subtle interaction, but it changes how the mode feels. WIDE’s M/S mode trusts that you have already addressed bass width in your mix and focuses on giving you clean, intuitive control over the stereo field.

full disclosure: i make KERN WIDE. the M/S mode described here uses the same encoding and decoding math as every other M/S tool. the difference is in the implementation details: how the Side boost interacts with the width shaping EQ, and how the correlation constraint prevents you from pushing into dangerous territory. try it on your mix and judge for yourself.

if you use M/S in a way i did not cover, or if something here does not match your experience, jonas@kernaudio.io.