What we mean by bright

Six adjectives, three frequency regions and one threshold, measured across a library of forty-eight records.

Ask three engineers what "warm" means and you get three answers, all confident. More low end. Less top. Valve distortion. A vague feeling about tape.

The word survives that vagueness because it is doing useful work — people genuinely hear the thing it points at — but a mastering tool cannot act on a feeling. So each of these adjectives here has a number behind it, and this note is the number.

The trick that makes it measurable

The first problem is that all music slopes downward. Every record has more energy at 80 Hz than at 16 kHz, usually by twenty or thirty decibels, because that is how instruments and rooms work. Measure a spectrum and call it dark because the top is lower than the bottom and you have described the entire history of recorded music.

So the words are not measured against flat. They are measured against what records typically do — a neutral curve built from the library itself. "Bright" means brighter than a typical record, not brighter than a straight line.

Three regions, each an average of three octave bands:

And one threshold, ±2.5 dB, applied to every region. Below that a record is just a record.

The six words

Two things follow from that table that are worth saying out loud.

Warm is a statement about treble, not bass. It is the absence of top, not the presence of bottom. A record can be warm and lean at the same time — quiet on top and light underneath — and several are. The intuition that warm means "more bass" is simply not what the ear is reporting.

Bass-heavy and lean are one axis, and warm and bright are another. They can combine in any of the four ways, and all four exist in the library.

Airy is not the same as bright

There is a seventh word, and it needed its own test.

The high region averages 5 kHz, 10 kHz and 16 kHz — and that average smooths away exactly what "airy" describes. A record can sit right at the norm across those three and still have a distinct lift in the very top octave, which is the sensation of air around a voice or a cymbal.

So airy is measured on the 16 kHz band alone, and at a higher bar: +4 dB. A record can be dark overall and still be airy. That combination sounds contradictory written down and is completely ordinary to listen to.

Where the library actually sits

The extremes, measured against the neutral curve:

Mothership Connection is the one to sit with. Eight decibels light in the low region and twelve decibels up in the top octave — a funk record that most people would describe from memory as thick and heavy, and which measures as the leanest and airiest thing in the collection. What reads as weight in that record is arrangement and playing, not spectrum.

The adjective describes the impression. The measurement describes the file. They disagree more often than is comfortable.

How this becomes an EQ

This is the part where the vocabulary earns its keep, and it is simpler than it sounds: the tonal EQ is the difference between two spectra, band by band.

gain at each band = (reference profile − your profile) × strength

Ten bands, ten subtractions. If a reference sits 4 dB above the norm at 10 kHz and your mix sits 1 dB below, the band gets +5 dB before the strength factor. Nobody types "bright" anywhere in that path.

Which raises the obvious question: what are the words for, then?

They are how you find the record. The matcher is textual, so airy and lean are how someone reaches Mothership Connection without knowing it is in the library. And they are how a preset can be checked: a record tagged warm that measures 8 dB above the norm on top has one of the two wrong, and that has happened — twice, to records whose reputations are older than the measurements.

The words that could not be measured

Some vocabulary was tried and dropped, and the pattern in what failed is instructive.

Punchy, open and crushed survive, but they are dynamics words rather than tonal ones, and they are measured completely differently — their own note covers why the two most obvious ones disagree with each other on a quarter of the library.

Muddy, harsh, boxy and thin did not survive. Each of them points at a narrow region — 200-400 Hz for muddy and boxy, 2-4 kHz for harsh — and an octave-band average is too coarse to see them. They are real and they are audible; they are just not visible at this resolution, and a tag that fires on the wrong thing is worse than no tag.

That gap is honest to state. Six words and one threshold do not describe a record. They describe the axes on which one record differs from another, which is a smaller claim and a true one.