The masked threshold is a measurement
Masking is not a mixing heuristic. It is a property of the ear with a measurement procedure. Fix a masker at a known level, play a probe tone at a fixed frequency, and lower the probe until the listener reports it is no longer there. That threshold rises as the masker gets louder and as the probe moves toward the masker in frequency. Repeat it across a grid and you get a two-dimensional surface, and that surface is the single most useful object in this whole area.
| Effect | When it applies | Practical consequence |
|---|---|---|
| Simultaneous masking | Both sounds at once | Dupes vanish inside one band |
| Release masking | Masker stops first | Piano notes reappear clearly |
| Backward masking | Masker starts after probe | Soft transients get swallowed |
| Inaudibility masking | Frequencies vanish | Phase tricks that erase a source |
The practical consequence is that clarity is mostly an arrangement question. Two sources inside the same critical band compete no matter how their levels are set, because you are asking one ear to resolve two things in one place. Moving one of them up an octave or several hundred hertz costs nothing and solves the problem permanently.
Critical bands and why they matter
The ear's frequency selectivity is described by critical bands, each roughly the bandwidth over which a masker and a probe interfere with each other. In the Bark scale used in the literature they number 24, running from about 100 Hz up to around 1550 Hz, and each band is wider than the last in absolute terms.
- Below about 500 Hz a band is roughly 100 Hz wide, so 80 Hz and 150 Hz are already audible as separate events.
- Between 1 and 3 kHz a band is a few hundred hertz wide, which is where most pop mix congestion lives.
- Above 5 kHz bands widen to around a kilohertz, so two cymbals competing up there will smear into one.
- Inside a band, two tones close together also beat against each other, which turns a static difference in frequency into a rhythmic pattern.
The often-repeated figure that two tones inside one band are indistinguishable is an oversimplification. Listeners can hear about a third of a critical bandwidth apart, and much less in the presence of noise, which is why doubling a guitar at the same octave is less muddy at the twelfth than at the unison.
Fletcher-Munson and level-dependent masking
The Fletcher-Munson curves, later revised as equal-loudness contours, map how the ear's sensitivity shifts with level. At low levels, hearing collapses above about 4 kHz and below about 100 Hz. As level rises, both ends open up, and at high levels the 4 kHz region becomes the most sensitive area of all.
This is why a mix judged on a laptop in a quiet room at low volume and the same mix judged on a system at high level are not the same mix perceptually. The top end may be inaudible in the first case, and the 2 to 5 kHz region may be exaggerated in the second. Neither judgement is wrong, and both are incomplete.
The trade-off with fixing this is the familiar one. Boosting presence to fix a low-level check will make the mix thin and fatiguing when played properly, and cutting presence to tame a loud playback removes the clarity that was missing. The honest move is to keep a fixed reference level for critical decisions and treat loud or quiet playback as a translation check.
Where the criticism lands
The claims worth pushing back on are the ones that promise to measure clarity. There is no agreed scalar for how intelligible a mix is, and loudness meters weighted by average power will report a bright thin master as louder than a balanced one, which tells you nothing about whether either is intelligible. A commercial plugin claiming to number clarity with a single dB figure is expressing an opinion.
What does survive is narrower and more useful. Masking curves predict what will disappear. Critical band widths predict which moves will be cheap. Equal-loudness contours predict how a judgement will change with level. None of those predict that a mix will feel good, and no instrument does.
The bottom line
Masking is the ear doing you a favour: one sound raises the threshold for another, and the size of that effect is predictable from critical bands and the Fletcher-Munson curve. Use it to justify a move you can hear and to predict which element you are about to erase, but stay away from level maps that claim to measure clarity.
Hear a critical band for yourself
Shift the whole project by one ratio with 432Hz MASTER and the band relationships move with it, so masking stays audible rather than becoming abstract.
Frequently asked questions
What exactly is auditory masking?
When one sound raises the hearing threshold for another at nearby frequencies, the second becomes harder to detect. It is completely measurable: play a probe tone at a given frequency and vary its level until it becomes audible again.
How wide is a critical band?
About 100 Hz near 500 Hz, growing to roughly a thousand hertz by 8 kHz, and narrower again at the very low end. The ear resolves frequencies inside one band less well than frequencies spaced across separate bands.
Does masking only happen between loud things?
No. A quieter sound can mask a louder one if it sits closer in frequency, and in a dense arrangement what disappears is rarely the loudest element. It is usually the one a few decibels down and a fraction of an octave away.
Can you un-mask after the fact with EQ?
Sometimes, and only in one direction. A narrow cut inside the masker's band can raise the masked source above the new threshold, but the cut is audible on its own and you cannot fix it globally without touching the wrong sounds.