The Fletcher-Munson Curve: Why Loud and Quiet Don't Sound the Same
Play the same mix quietly and it feels thin; turn it up and the bass fills back in - the ear's sensitivity curve, not the mix, is what changed.
In 1933, Bell Labs researchers Harvey Fletcher and Wilden Munson measured how loud a given frequency needs to be, in decibels, to sound as loud as a 1kHz reference tone to the average listener. Plotted across the audible range at different volume levels, the results became the equal-loudness contours now generally known as the Fletcher-Munson curves - later refined by Robinson and Dadson in 1956 and formalized as the ISO 226 standard still used today.
The core finding: the ear's sensitivity is not flat across frequency, and how uneven it is depends on how loud the sound is overall. At low listening levels, bass and the very top of the treble range both need considerably more sound pressure to register as loud as a midrange tone - the ear is most sensitive between roughly 2kHz and 5kHz, the range that carries most of human speech. Turn the volume up, and the contours flatten out; low and high frequencies close the gap with the midrange.
That has a direct consequence for mixing and monitoring: a mix judged 'balanced' at a quiet listening level will typically sound bass-light and treble-light once played back louder, and a mix balanced loud will sound thin and boxy when played quiet. It's the reason professional rooms monitor at a calibrated, consistent SPL rather than 'whatever sounds right in the moment,' and why consumer hi-fi gear often includes a 'loudness' button that boosts bass and treble specifically at low volumes to compensate for the ear's own curve.
It also matters outside the studio. In a venue running background music, staff routinely turn the volume up and down through the day without touching the EQ - but because the Fletcher-Munson curve shifts with level, the perceived tonal balance changes at every volume step even though the DSP settings never do. A curve that reads flat at one SPL won't necessarily read flat at another, which is one more reason a system's EQ needs to be tuned with the room's actual working volume in mind, not just its loudest moment.
Based on Fletcher and Munson's original 1933 equal-loudness research (later refined as ISO 226) - covered in Philipp's psychoacoustics coursework at BIMM Berlin.