By the end: the two numbers that decide whether your wide parts survive a mono system — stereo correlation and fold loss — and how to measure both on your own tracks before a club rig measures them for you.
Your mix sounds wide and expensive on headphones. Then it plays somewhere that sums to mono — a club rig, a car, a phone speaker — and the atmosphere you spent an hour building is suddenly quieter, thinner, or simply gone. That is not taste and it is not bad luck. It is arithmetic, and you can run it in your DAW in about a minute.
Step 1
Club systems sum the bottom end to one stack of subs, and plenty sum more than that. Cars and phones sum further still. Anything in your mix whose left and right channels disagree gets subtracted at the sum — the wider and more decorrelated a part is, the more of it the room never hears. Width is a bet that the listener hears two channels. The fold test prices that bet.
Step 2
Solo the part in question — a pad, a bed, a wide synth — and sum it to mono: a utility mono switch, a channel sum, whatever your DAW gives you. Listen first. Then measure the folded version against the stereo original — same section, same gain. Two numbers tell the whole story.
Step 3
Correlation asks whether the channels agree: +1 means identical, 0 means unrelated, and negative means they are actively fighting — those are the parts mono deletes. Fold loss asks what the sum costs: the folded level against the average of the two channels, in dB. Together they separate “wide” from “broken” — a part can be wide and fold clean, or narrow and still cancel.
Step 4
This test exists because our own files failed it. CONCRETE’s atmosphere beds were built wide — and when we folded them, the width turned out to be cancellation wearing a costume. Emin_06, the worst of the three, summed to almost nothing: correlation −0.431, fold loss 5.43 dB. The fix re-phased the channels without touching the music — same notes, same frames, loudness within 0.09 LUFS — and the re-measure cleared both bars on all three beds.
Step 5
Work in this order: mono everything below ~120 Hz (that law is Mistake 4 of the low-end diagnostic); move the width up into the top layer, where cancellation costs less; and if one element is generating the negative correlation — a chorus, a widener, a doubled take — re-phase or re-pan it at the source. Then re-measure. A fix counts only when the numbers move; a fold you didn’t re-measure is a hope, not a fix.
Check your work
Level does not repair cancellation. A part at −0.43 correlation, folded louder, is a louder absence — the sum still subtracts it. Second failure: folding only the full mix. A mix can pass on average while one bed is being eaten whole. Fold per part; the one that vanishes solo is the one the rig eats.
The beds in this lesson ship in CONCRETE — six 8-bar atmosphere beds, key-labeled. Emin_06 is in the pack exactly as it was measured above: run the fold on it and check our arithmetic.
Hear it in a pack
Open the pack for the beds’ full demo list, the contents, and the measured facts behind them.
▶ Start here
Peak-time techno foundations: driving drums, rolling bass, dub stabs and atmosphere — every file measured, key-checked and ear-checked before it listed.
Lab24-bit / 44.1 kHz · Peak ceiling −1.0 dBFS · Keys measured vs labels
How-To · Related
Mistake 4 is this test’s doorway — the stereo sub. The other four construction errors are in the same family: fix the build, not the mix.
Read the how-toHow-To · Related
The other measurement your meter hides from you: the peak between the samples. Same doctrine — measure it, then ship to the number.
Read the how-toStay in the loop
New packs and free drops, first — follow Modern Tastemaker on Gumroad and they land in your inbox the day they ship.
Double opt-in — Gumroad emails you one confirm link. Unsubscribe anytime.
