Weighting networks and Leq: two conventions that decide the number

A sound level meter does not report pressure. It reports pressure passed through a filter chosen by a committee in the 1930s, and the choice changes the answer by tens of decibels at the ends of the band. Two meters set to A and C can read the same event 25 dB apart and both be right.

The second convention is how the reading is averaged over time. Leq is an energy average, which means a brief loud event dominates it far more than its duration suggests — and that is the number nearly every noise standard is written against.

What the networks are

All three are filters applied to the pressure signal before it is squared and averaged. Z is the flat one and is what "dB SPL" means when nobody names a network. C is close to flat across the audio band, rolling off gently below 30 Hz. A is the one with the shape everyone recognises: nearly flat from 500 Hz to 5 kHz, falling steeply at both ends.

The curves come from equal-loudness contours, which are themselves averaged over many listeners at a defined level. A follows the 40-phon contour — the shape of equal loudness for quiet sounds — and C follows the 100-phon contour, which is much flatter because loud low frequencies are heard more efficiently.

The A-weighting is a convention, not a correction

It is worth being precise about what the network does. It does not restore what you hear; it reports what the sound would be worth if it were quiet. At 50 Hz it reads 30 dB low, at 100 Hz 19 dB low, and at 1 kHz by definition zero.

That is why a mix can measure comfortably under a limit and still be dominated by a 45 Hz room boom: the meter has been told to ignore most of it. Using A-weighting to judge bass is using a filter designed to discard the evidence.

Why meters default to A anyway

Because hearing damage and annoyance correlate better with dB(A) than with dB(Z) for most industrial and environmental noise, which is mid-band dominated. Every occupational limit, every planning condition and most product specifications are written in dB(A), so the convention is self-sustaining regardless of its suitability for music.

The practical rule for audio work: use A when you must report against a standard, and use C or Z when you are diagnosing a room. Report both if either audience might read it.

Leq averages energy, not level

Equivalent continuous level is the steady level that would deliver the same total energy over the same time. Because decibels are logarithmic, the arithmetic happens on linear energy and only the result is converted back.

The example in the calculator makes the point: 100 dB for one minute plus 70 dB for nine comes out at 90 dB, not the 73 dB an average of the levels would give. The loud minute carries a thousand times the energy of each quiet one, so it sets the answer.

This is also why an impulsive event is so damaging and so hard to describe: a single gunshot or snare hit can carry more energy than an hour of the background it interrupts, and a one-second Leq will say so while a peak meter says nothing about duration at all.

Dose and the 3 dB exchange rate

Occupational limits are expressed as a dose over a working day, and the exchange rate says how much the allowed time halves for each increase in level. The equal-energy rate is 3 dB: 85 dB(A) for eight hours, 88 dB(A) for four, 91 dB(A) for two.

Two consequences follow that surprise people. A 3 dB change is a doubling of energy, not a small step. And at 100 dB(A) the permitted exposure is fifteen minutes, which is roughly one loud rehearsal — the arithmetic is unforgiving because it is an average of energy, not of comfort.

Five mistakes worth avoiding

1. Averaging decibels directly. Two 80 dB events for equal times give 83 dB, never 80.
2. Quoting dB(A) for bass problems. The network removes exactly the evidence you are trying to look at.
3. Treating Z as "true". It is flat, which is a different convention, not a more honest one.
4. Comparing Leq with peak. They answer different questions — one is energy over time, the other is the largest instantaneous excursion.
5. Forgetting that Leq depends on the window. The same event gives a different Leq over one second and over one hour, and both are correct.

Summary

Weighting is a filter with a history; Leq is an energy average with a window. Neither is more true than the other, and both are decided by convention. State which network and which window you used, or the number is not reproducible.

Values shown are engineering aids rather than measurements; see the disclaimer and the tool index.