Equal Loudness Curves

Goal

Understand the Equal Loudness or Fletcher Munson curves.

What is loudness?

Loudness is something that is determined by humans. Notice we do not measure loudness, we measure pressure, or intensity, or something else that can be directly measured. What is loud to one person may be soft to another. For example, you may find a loud 10 kHz tone very annoying but an old 80 year old man may barely notice it!

We do have a "loudness unit" now, called LUFS, or Loudness Unit Full Scale. However, how we arrived at such a unit is what we will be more concerned with here. How does one begin to quantify something like "loudness"? Well, Harvey Fletcher and Wilden A. Munson decided to do a simple test.

The idea of a loudness curve

While loudness may be hard to pin down, perhaps variation in loudness could be more predictable. Here is the idea, we play a specific reference frequency at a specific SPL. We then play another "test" frequency at that exact same SPL. We then ask the listener if the test frequency should be turned up or down to match the SPL of the reference tone. We then continue to do this until the listener has concluded they are the same loudness. The difference in movement is thus what it take to make that test frequency the same "loudness" as the first.

We can continue to do this for all frequencies at that particular SPL until we have done them all. It will produce a curve known as an "equal loudness" curve because it is how much each frequency must be turned up or down to get that loudness.

Example

To make it a little more concrete suppose we ran this test. We decided to choose a reference frequency of 1 kHz at 40 dBSPL. For our test subject we play the 1 kHz tone at 40 dBSPL. We then play 100 Hz at 40 dBSPL. We then ask them if it should be turned up, left alone, or turned down to match the reference. We receive the answer to turn it up. We do so and continue to ask until they say they are the same. In the end the 100 Hz signal is now at 57.7 dBSPL. That is to say, if we want 100 Hz to sound sound the same as 1 kHz at 40 dBSPL we would need to turn it up by 17.7 dBSPL.

Contour Lines

We can then take this a step further. Say our first run we used 1 kHz as the reference tone and ran the test at 40 dBSPL. Now suppose we ran it again for a reference of 50 dBSPL, and then 60 dBSPL etc... Each time we would generate an entire new "equal loudness" line.

These experiments have been repeated, with the most recent standard of ISO (International Organization for Standardization) 226 being done in 2023. Most current charts show the original Fletcher Munson curves (1933) over layed with the 2003 revision. Note that these curves are averages, your specific curve will be different. Also keep in mind that the number of frequencies tested will affect the data as well.

Because we use the more recent measurements for our standards the name "Equal Loudness Contour" is a better fit than "Fletcher Munson Curves" though you will often see the names interchanged freely.

Equal Loudness Contours


Observations

First note that at low dBSPL we need to turn up low frequencies a lot, and high frequencies as well. The lower the frequency the more it must be turned up.

Next note that at as dBSPL increases, the curve tends to flatten out suggesting that higher dBSPL yields a more "even" representation of the spectrum to us. The resonance of the ear also tends to play less of a role as the dBSPL increases.

Is it loudness?

While this shows there is a spectral imbalance that appears to be consistent across people, consistent enough to produce these averages, we still have several other factors to consider. The length of a sound, the content of a sound. A sound amount of "change" is has spectrally. These are major factors in loudness not considered here. As such while the equal loudness contours give us some general hints it is clear they fall short of "loudness" in the broad sense of the word.

For the remainder of this lesson when I use the term "loudness" just know it comes with all of these conditions. It's not true "loudness" but it is much better than just using dB SPL as that is clearly way off if such large adjustments are need to make things sound "equally loud"!

The Phon

We need to move away from the dB SPL when talking about a specific loudness line. For example it is a mouthful to say 100 Hz with the reference of 1 kHz at 40 dBSPL. So what we do instead is define the reference to be 40 phons. Then every point on that line is understood to be referenced to the reference at 40 dBSPL.

The phon thus relates the loudness to the dBSPL, but it gets rid of the need to constantly state the reference. You can say the 40 phon line at 100 Hz or 10 kHz and it is understood that it meant to use the reference of 1 kHz at 40 dBSPL. More importantly the change in dB SPL is included in the phon line directly. We don't need to state how much 100 Hz had to change in dBSPL because we are talking about the specific loudness line. It is understood you would have to look it up.

The Sone

An increase of 10 dBSPL is associated with a perceptual doubling of loudness, as a kind of quick and dirty rule. The problem with this is using dBSPL for loudness is not a great measurement. We have related dBSPL to loudness through the phon. This does take into account the variation required to keep the loudness equal, hence going up or down by 10 phons is similar to doubling or halving the loudness. We define the reference (1k at 40 phon) to be 1 sone. Then for a 10 phon increase the sone will double, and for a 10 phone decrease the sone will half. This gives the following scale:

Sone Phon

.25

20

.5

30

1

40

2

50

4

60

8

70

16

80

32

90

64

100

So if we describe the loudness in sones a doubling of sones directly relates to a "doubling" of loudness. This scale is for sine waves, there are nuances I am leaving out for more complex signals. I just want you to know what the "sone" scale is.

Questions

  1. How does the phon scale differ from dBSPL?

  2. What would be the phon rating of a 100 Hz sound that has a decibel rating of 90dBSPL?

  3. How is a sone different than a phon?

  4. How many times louder is an 80 phon sound to a 40 phon sound? (Hint: use sones.)

  5. How many times louder is an 60 phon sound to a 70 phon sound? (Hint: use sones.)

  6. A sound is measured to have 90dBSPL at 100 Hz. What is its sone?

  7. A sound is measured to have

    104Wm2 10^{-4}\frac{W}{m^{2}}
    at 100 Hz. What is its sone?

To support this series please consider donating via

paypal

or joining the

patreon.