Turn up the width on a finished mix and it sounds better immediately. It sounds bigger, more open, more expensive. This happens every time, on every mix, including the ones that were already too wide.
The reason is that widening raises the side channel, and raising the side channel makes the mix louder overall while moving energy away from the centre where the ear expects density. Both of those read as improvement in the first few seconds. Neither survives a fold to mono or a second listen an hour later.
A master is the wrong place to find out which one you did.
What “wider” actually does
Every width control does the same underlying thing: it changes the ratio between the mid channel and the side channel. Above 100% the side gets louder relative to the mid. Below it, quieter.
Everything else follows from that. A wider master has more energy in the difference signal and less in the sum, which means:
- The centre gets relatively quieter. Vocal, kick and bass lose density.
- The mono fold loses more, because side content is what cancels.
- The perceived level goes up in stereo and down in mono, so the same master reads at two different loudnesses depending on the system.
None of that makes widening wrong. It makes it a trade, and a trade is something you measure.
The three things to check
Correlation. A full mix normally sits somewhere between +0.3 and +0.7. Pushing a master below +0.2 is a lot of width for finished material. Below zero for sustained passages means content is cancelling rather than spreading.
Mono loss per band. The number that tells you what the fold costs. Under 1 dB across the bands is comfortable. Three is noticeable. Six in any band, and especially the low one, means the master is built on width that half your listeners won’t hear.
Width against the reference. Load two or three commercially released tracks in the same style and read the same numbers off them. This is the only calibration that means anything, because “wide” is genre-relative. A modern pop master and an acoustic record sit in different places and both are correct.
If your master reads wider than every reference you trust, the question isn’t whether it sounds good. It’s why nobody else is doing it.
How much a master should add
Usually less than people expect.
A mix that arrives well balanced needs no width at all. The stereo image is a mix decision, made with panning and with the reverb sends, and a mastering-stage width control is a blunt instrument by comparison. It moves everything at once, including the things that were already where you wanted them.
Where a small amount is worth it:
A mix that came in narrow. Five to ten percent, full band, is a real fix and nobody will hear it as processing.
A mix that’s wide up top and narrow down low. This wants a band-limited move, not a full-band one. Add width above 300 Hz and leave the bass alone.
A mix where the sides are dull. Reach for a high shelf on the side channel rather than a width control. It gets you the openness without the level shift, and it doesn’t cost you anything on the fold. See how to use mid/side EQ.
Where it isn’t worth it: as a finishing move applied to everything by habit, or to make a master compete with a reference that’s louder rather than wider. The second one is a level problem being solved with the wrong control.
The order to work in
Width interacts with everything after it. Widening before a clipper changes what the clipper catches, because the side channel’s peaks move. Widening after a limiter undoes some of the level control you just set.
Put the width move early in the chain, before dynamics, and set it once. Then leave it alone while you work on level, and check it again at the end against the same three measurements.
A starting point
- Read correlation and per-band mono loss on the unprocessed mix. Write the numbers down.
- Read the same numbers on two references.
- If the mix is already in the same range, don’t add width. This is the most common correct answer.
- If you do add it, add it band-limited above the bass and in single-digit percentages.
- Fold to mono at matched level and confirm you didn’t lose more than a decibel. If you did, take the width back until you didn’t.