Monitoring level changes the balance you hear

The late-night low end

You finish a mix in a flat with thin walls, speakers turned well down so the neighbours do not knock. The kick and bass feel a little thin, so you lift them until they feel right. The next afternoon you play it at a normal level and the low end is huge, and in the car it is worse.

You balanced the low end at a very low playback level, where your hearing gives bass less weight than it does at a normal level. The bass you added to make up for that is too much once the volume comes up.

Why it matters

Every EQ move goes through your ears at one level

You never hear a mix's frequency balance directly. You hear it through ears whose sensitivity to bass, mids and top changes with how loud the playback is. Make all your low-end calls at a low level and you add bass the mix does not need. Make them all loud and you tend to hold the bass back, as the lesson on loudness bias explains. Swing the volume knob between the two without noticing and every comparison you make is a little rigged.

The science

Equal-loudness contours

An equal-loudness contour shows the sound pressure level a pure tone needs at each frequency to sound as loud as a 1 kHz tone. The level of that 1 kHz tone names the contour in phon, so the 60 phon contour passes through 60 dB at 1 kHz. ISO 226:2023 gives a family of these contours for young adults (18 to 25) with normal hearing, listening to tones from the front in a free field. Suzuki, Takeshima and Kurakata (2024, p. 5) compared the two editions at every third-octave frequency in 10 phon steps: the differences stay within 0.6 dB, and within 0.3 dB above 10 phon. The numbers below are therefore worked out from the published 2003 equations.

On the usual chart, drawn in absolute level, the contours crowd together in the bass: going from 40 to 80 phon takes 40 dB at 1 kHz but only about 24 dB at 50 Hz. Drawn relative to 1 kHz, as here, the same fact shows up as curves that fan apart toward the low end. To sound as loud as the 1 kHz tone, a 50 Hz tone needs 21.7 dB more at 80 phon, 30.0 dB more at 60 phon and 37.8 dB more at 40 phon. Turn the playback down from about 80 to about 40 phon and 50 Hz falls behind the midrange by about 16 dB. At 100 Hz the shift is about 12 dB.

The top end moves much less. At 8 kHz the curves sit within half a decibel of each other across those levels. Only near 12.5 kHz, the top of the standard's range, do they spread again, by about 6 dB from 40 to 80 phon, which is the air you lose when you turn down.

Two limits keep this honest. The contours were measured with pure tones, and a mix is broadband sound where parts mask each other, so the numbers show the direction and rough size of the shift, not an EQ correction to apply. People also differ: the experimental equal-loudness data behind the standard typically scatter with a standard deviation of about 5 to 6 dB (Suzuki et al., 2024, p. 4).

Play the same mix at three levels and listen to how much low end and air you hear at each one.

A repeatable level keeps your comparisons fair. Katz (2000) calibrates each speaker so pink noise at -20 dBFS RMS reads 83 dB SPL, C-weighted, slow, at the listening position, and turns the gain down from that reference for more compressed material. You do not need his number to get the benefit. You need one level you can come back to. Keep the loud checks short; the lesson on fresh ears has the exposure limits.