Reading a correlation meter: one number, four bands, and the mono truth

Club rigs, phone speakers, Bluetooth boxes, and plenty of broadcast chains sum your stereo mix to mono without asking. Whatever the left and right channels disagree about gets weaker in that sum — and if they fully oppose, it vanishes. Correlation is the number that predicts the damage before a sound system demonstrates it.

This page reads that number for the whole file, then goes further than the classic meter: per-band verdicts, an audible rendered fold-down, and a side-by-side check for two masters.

§How to check a mix's mono compatibility

  1. Drop the mixThe whole file is scanned locally; flagged sections appear on the waveform in seconds.
  2. Read the verdictOverall correlation plus the zones where it collapses. Play the track and watch the scope ride along.
  3. Find the guilty bandOpen By band — four correlation readings localize whether the risk is bass, mids, or air.
  4. Hear the fold-downOpen Mono fold-down, render the sum, and read the loss report next to what your ears report.
  5. Fix and re-checkNarrow the widener, mono the lows, re-drop the file. The meter is only useful in a loop.

01What correlation actually measures

The meter multiplies the left and right channels sample by sample over a short window, then normalizes by each channel's energy. Identical channels score +1, unrelated ones score near 0, and opposite-polarity channels score −1. Since mono is the average of the two, a −1 pair cancels to silence in the fold — the meter is literally a prediction of that subtraction.

Real mixes breathe between about +0.3 and +1. Zero is not failure — it just means wide — but sustained negative territory means the channels are fighting, and mono is where the fight ends badly.

02One number can lie; four lie less

A whole-file reading blends everything together, and loud mono elements mask quiet phase problems. A dead-center kick can hold the overall number at +0.55 while an out-of-phase synth pad quietly dies underneath it. The By band mode splits the file into lows, low-mids, highs, and air, and scores each on its own.

Worked example: overall +0.55 looks healthy, but the 120–600 Hz band reads −0.2 while carrying a third of the mix's energy — that band is the actual problem, and it points straight at whatever stereo-widened instrument lives there. The verdict chips weight each band's reading by its energy share, because a cancelling band that holds nothing matters to nobody.

03Don't trust the picture — render the sum

Meters predict; the fold-down proves. The Mono fold-down mode renders (L + R) ÷ 2 as a real WAV you can play and keep, next to a loss report: integrated LUFS of the stereo file versus the mono render, and the energy lost per band. A drop of a fraction of a dB is a healthy mix; a drop of several dB is width built from phase tricks, now collected.

Listening to the render matters more than the numbers — a mix can lose a dB of measurable energy and still groove, or lose the one synth hook that mattered.

04Two masters, one table

The question is rarely 'is this mix mono-safe' in a vacuum — it is 'is my master behaving like the reference'. Compare two takes a second file and lays the numbers side by side: correlation, width, mono loss, each with its own player so you can flip between them by ear.

If your master shows twice the mono loss of the record you are chasing, the difference is not taste — it is phase, and now it has a number.

05What the meter cannot tell you

Correlation measures how much the channels agree, never which instrument makes them disagree — that identification is your ear's job, with solo buttons or stems. Brief negative flickers on cymbals and reverb tails are normal physics of wide, dense material; the meter only convicts on sustained negative readings. And a perfectly mono file scores +1 while sounding narrow as a hallway — the meter grades safety, not beauty.

06Meters diagnose. Stems operate.

When the verdict says your low end cancels, the fix needs the parts, not the picture — you have to reach the bass itself. SongStrip in the studio runs GPU stem separation that knows what the instruments are and hands you the vocal, drums, bass, and music as clean stems, so the out-of-phase element can be rebalanced at the source instead of EQ'd around. Diagnose here; operate there.

Take it further in the studio — start free with 3 full packs, no card needed.