Mono is not a historical problem. A phone speaker is mono. A laptop’s two drivers are close enough together to be mono at low frequencies. A club system usually runs the subs in mono on purpose. A voice assistant, a TV, a shop ceiling, a car with one blown door speaker. All mono, and all places your track will be heard.
When a stereo file folds to mono, the two channels sum. Anything the speakers play equally gets louder by 6 dB relative to itself. Anything they play in opposite polarity cancels and disappears. Most content sits between those two extremes and loses some amount of level in the fold.
The problem is that the loss isn’t spread evenly. It concentrates in whatever you widened.
What cancellation costs
Two identical signals in opposite polarity sum to silence. That’s the extreme case and it’s rare. What’s common is partial cancellation: a wide synth loses 3 dB, a reverb loses 4, a doubled guitar loses 2, and the mix that felt full in stereo comes out thin and small on the fold.
The frequencies where it hurts most are the low ones. Bass carries the most energy, so a few decibels lost below 150 Hz is a bigger change to the record than the same loss at 8 kHz. It’s also the range where a stereo widener does the most damage, because widening moves energy out of the centre where the bass lives.
Reading correlation
A correlation meter reads from +1 to -1.
+1 means the two channels are identical. Mono content. It folds perfectly and gains 6 dB.
0 means they’re unrelated. This is normal and fine for wide stereo material. A fold costs you around 3 dB on that content.
-1 means they’re identical and opposite. It cancels completely on a fold.
Anything sitting below zero for a sustained stretch is worth investigating. Brief dips are normal: a reverb tail, a wide transient. A meter parked at -0.3 while the chorus plays means something in that chorus is going to vanish.
The catch is that one number across the whole spectrum tells you very little. A mix can read +0.4 overall and still lose 5 dB of bass, because the bass is a small part of what a full-band correlation meter is looking at. Per-band correlation is what makes the reading actionable.
Mono loss is the number that matters
Correlation tells you how related the channels are. It doesn’t tell you what a fold will cost, and that’s the question you’re actually asking.
Mono loss does: it’s the difference in decibels between the level now and the level after summing. Minus one decibel is nothing. Minus three is audible. Minus six on the low band means half the weight of your record is going to disappear on a phone, and you want to know that before you print.
Read it per band. A mix that loses 1 dB up top and 6 dB at the bottom needs a completely different fix from one that loses 4 dB evenly.
Why the low end needs a longer look
Frequency resolution and time resolution trade against each other. A normal analysis frame — around 2048 samples — resolves the top of the spectrum in plenty of detail and leaves the bottom two octaves as a handful of blurred bins. Below 200 Hz you might have eight bins covering the entire range where your kick, your bass and your sub all live.
Eight bins can’t tell you that the trouble is at 60 Hz and not at 110. A frame four times longer gets you ten third-octave rows across the same range, built from around seventy bins, which is enough to name the frequency. It updates more slowly, which is a fair trade for a measurement you take deliberately rather than watch continuously.
The useful output isn’t a needle. It’s a sentence: the worst band is 63 Hz, a fold costs you 5.2 dB there, and mono below 120 Hz gets 4 of it back.
The fixes, in order
Mono the low end. A crossover that sums everything below 100–150 Hz to the centre solves most bass fold problems in one move and is inaudible in stereo, because you couldn’t localise those frequencies anyway.
Find the polarity problem. If a whole element cancels rather than thins out, one channel is flipped somewhere. Check doubled takes, sampled stereo one-shots, and anything that came out of a stereo widener with the width past 150%.
Back off the widening. Width plugins that work by inverting and delaying a channel buy stereo spread with mono compatibility. If a fold costs you 5 dB and you widened three things, undo the widening before you reach for anything cleverer.
Narrow the reverb, don’t remove it. Reverb returns are usually the largest side-channel contributor and the easiest to pull in slightly without anyone noticing in stereo.
A starting point
- Check mono at the end of the mix, not the end of the master. The fixes live in the mix.
- Look at the per-band mono loss before you listen. Knowing the number stops you talking yourself into a fold being fine.
- Fold to mono and listen once at the same level. Not louder, not quieter.
- Fix the low band first. It’s the largest number and the cheapest fix.
- Re-check. Every fix moves the others, and one crossover often takes care of the whole thing.