Turn the monitor knob down and the kick drum thins out before anything else does. Turn it back up and the bottom returns. The bass did not actually move. Your hearing did.
This is the practical face of one of the most studied effects in audio: the human ear is not a flat microphone. Its sensitivity depends on both frequency and absolute level, so the same mix bus sounds like a different EQ at 65 dB SPL and at 85 dB SPL. The point of this article is not to repeat that as a slogan. The point is to look at what the measurements actually say, then use those numbers to set a sensible monitoring level you can return to every session.
What Fletcher and Munson actually measured
In 1933, Harvey Fletcher and Wilden Munson published "Loudness, Its Definition, Measurement and Calculation" in the Bell System Technical Journal (Vol 12, Issue 4, pp. 377-430). Both worked at Bell Labs. Their experiment was deceptively simple: subjects wore headphones and matched the loudness of pure test tones to a 1000 Hz reference, sweeping frequency and intensity in 10 dB steps. The resulting family of curves connects every (frequency, SPL) pair that sounds equally loud as the 1 kHz reference at a given level.
The unit is the phon. A 1 kHz tone at 70 dB SPL is by definition 70 phons. The 70-phon contour traces the SPL that every other frequency needs to match that perceived loudness. Two features matter for mixing.
First, the curves are not flat. Hearing peaks in sensitivity around 2 to 5 kHz, driven mostly by the resonance of the ear canal (around 2.5 cm long). It rolls off steeply below 200 Hz and above 10 kHz.
Second, the curves are level dependent. At 80 phons the contour is comparatively flat. At 20 phons it has a deep bass shelf. A concrete number that lands the point: at 20 phons, a 100 Hz tone needs roughly 45 dB SPL to sound as loud as a 1 kHz tone at 20 dB SPL. That is about a 25 dB difference. At 80 phons the same comparison only needs about 10 dB more SPL at 100 Hz. The bass perception gap closes as you turn up.
The history is messier than the textbooks suggest
The Fletcher-Munson curves were the reference for two decades. Then in 1956 D.W. Robinson and R.S. Dadson re-measured them in free field, using loudspeakers in an anechoic chamber rather than headphones. Their data became the basis of the original ISO 226, published in 1961, and the audio industry treated those curves as canonical for over forty years.
When ISO 226 was rebuilt in 2003, the working group pooled twelve coordinated studies across Japan, Germany, Denmark, the UK and the US. The result was awkward: Robinson-Dadson 1956 was the outlier. Fletcher and Munson 1933 sat much closer to the modern measurements than the curves that had replaced them. Two generations of engineers had been taught to correct against the wrong standard.
ISO 226:2023 (Edition 3, published March 2023) is the current revision. Practically, it sits within 0.6 dB of ISO 226:2003 across the contours. The main changes are a power-law loudness exponent drawn from research by Suzuki and Takeshima in the mid-2000s, and a 0.4 dB downward adjustment of the 20 Hz hearing threshold to align with ISO 389-7:2019. For mixing decisions the 2003 and 2023 numbers are interchangeable.
Why this turns monitor volume into a mix decision
The level dependence is the part that matters at the desk. When you turn the monitor knob down, you do not just hear the same balance at a lower volume. You slide down the family of curves, and the bass shelf gets steeper. Less perceived bass is not a property of the mix. It is a property of you, at that SPL, in that moment.
This is why 1970s hi-fi amplifiers had a "loudness" button. The contour switch tapped a frequency-dependent network into the volume path that boosted bass and treble more as the volume dropped, approximating the spacing between adjacent phon contours in reverse. It is an approximation rather than a correction, because the amplifier does not measure the actual SPL at your ears.
The same logic argues against weighting filters that themselves embed the curves. A-weighting is built from an inverted 40-phon contour. Using A-weighted SPL to calibrate a monitoring level would double count the very effect you are trying to design around. For monitor calibration the convention is C-weighted, which is effectively flat from 31 Hz to 8 kHz with mild rolloff outside that band.
Picking a reference level
The defense against level-dependent perception is a calibrated monitor reference. Bob Katz's K-System, published in a 2000 AES paper and expanded on digido.com, is the most common framework in mastering. The procedure:
- Play -20 dBFS RMS pink noise from one channel at a time.
- Measure SPL at the listening position, C-weighted, slow response.
- Adjust the monitor controller gain until you hit a target SPL per channel.
- Mark the knob position. That is your reference.
Katz uses 83 dB SPL for mastering work in a treated room, sitting at the flatter end of the equal-loudness contours. Cinema uses 85 dB SPL per channel under SMPTE RP 200, a number driven by Ioan Allen's work at Dolby starting in the mid-1970s. Allen's original motivation was tying calibration to 0 VU on magnetic film. The 83 dB figure is Katz's choice, not Allen's, and conflating those two attributions is one of the most common mistakes online.
Smaller rooms with near-field monitors usually calibrate quieter. The Katz K-System defines a pop-music reference 6 dB below the film/mastering reference, around 77 dB SPL per channel. Sound on Sound and other practitioner sources cluster project-studio recommendations around 76 to 79 dB SPL. The exact number matters less than the discipline of having one and returning to it.
The workflow that follows from this:
- Make critical EQ and balance decisions at your calibrated level.
- Use lower levels, mono summing and a single small speaker as cross-check passes, not as the primary mix view. Those test how the mix translates, not how it should be balanced.
- Stop chasing "add more bass" decisions made at low monitor volume. At 60 dB SPL the ear is genuinely 15 to 25 dB less sensitive at 60 Hz than at 1 kHz. Mixes shaped to compensate for that perception will sound boomy on anyone else's playback.
A C-weighted SPL meter or a calibrated phone app (Decibel X, NIOSH Sound Level Meter, Sound Analyzer) is enough for setting the reference. A bandlimited -20 dBFS RMS pink noise file is the standard test signal, and most mixing tutorial sites host one.
The hearing-safety reality
A working day at 83 to 85 dB SPL is not free. NIOSH's recommended exposure limit is 85 dBA for 8 hours, with a 3 dB exchange rate, meaning every additional 3 dBA halves the safe time. OSHA's enforceable Permissible Exposure Limit is 90 dBA for 8 hours with a 5 dB exchange rate, slightly more permissive but covering the same ground.
C-weighted SPL on full-range program tends to read slightly higher than A-weighted SPL because the low end is no longer rolled off. The exact relationship depends on the spectrum. Practically: full days at K-System reference are at the edge of NIOSH's daily budget. The right response is to keep critical-decision passes short, drop the level for routine editing and comping, and take real breaks. The ear recovers from short loud exposure better than from extended moderate exposure.
What to take away
Three things are worth holding onto from this. The Fletcher-Munson contours are not abstract psychoacoustics, they are the reason your low-end perception changes when the volume knob moves. Picking a monitor reference SPL and sticking to it is the cheapest mix improvement most home studios could make. And the textbook history is more interesting than it looks: the curves your tutorials reproduce are 93 years old, and they turned out to be closer to current data than the curves that replaced them in 1961.
The mix knob you have not been thinking about as a mix decision is the monitor knob.