One Number Can Lie: Reading Stereo Correlation by Band
A stereo correlation meter gives you a single number, and it's tempting to treat that number as a pass or fail. Don't. A track can read as perfectly mono-safe on a full-band meter while still hiding a real problem underneath, or read as narrow when what's actually happening is something else entirely. The fix is checking correlation by frequency band instead of trusting the one overall reading.
Why the single number misleads
Correlation measured across the whole spectrum at once averages everything together. A track with a canceling bass problem and an unusually wide top end can average out to something that looks acceptable, even though the bottom end genuinely falls apart in mono. The reverse happens too, a track can feel narrow on first listen when the actual cause is a dark top end, nothing to do with width at all.
The only way to know which is happening is to stop measuring the whole mix as one number and start checking specific ranges separately.
The gradient that actually works
Bass, below roughly 150 Hz, should sit close to fully correlated, close to 1.0. This is what keeps the low end intact on a phone speaker or a mono Bluetooth speaker instead of thinning out or partially cancelling. If your bass isn't close to mono-safe, that's worth fixing before anything else, since it's the range most likely to cause real problems on real playback systems.
Mids can open up moderately, something in the range of 0.5 to 0.6 correlation is common and genre-appropriate for a lot of modern production. This is where a mix starts to feel like it has real space without risking translation problems.
Highs can go wider still, correlation in the 0.3 to 0.5 range at the top end is often exactly right, especially in polished, glossy genres where air and shimmer benefit from real stereo spread.
What this actually catches
Two different tracks can produce the same overall correlation reading for completely different reasons, one because it's genuinely well-built across the whole spectrum, the other because a mono-unsafe bass problem happens to average out against an unusually wide top end. Only checking by band tells you which one you're looking at.
It also resolves a specific, common confusion: a mix that feels narrow isn't always a width problem. If the bass and mids check out but the top end is dark or restrained, that missing brightness can read as narrowness even though the actual stereo image is fine. Fixing tone in that case does more than fixing width ever would.
How to check it
Most metering plugins let you filter or solo a frequency range before reading correlation, or you can use a multiband analyzer that shows correlation per band directly. Check bass, mids, and highs separately rather than reading the single full-band number and calling it done. It takes an extra minute and it's the difference between actually knowing what your stereo image is doing and guessing from one number that might be hiding the real story.
— Josh Founder, JG BeatsLab
