Perception of loudness

Now that we’ve talked about the physics of spectral distribution a little bit, we’ve got to discuss how we perceive loudness across the spectrum. There are two primary things to keep in mind: first, our hearing is most sensitive around 4 KHz. (Octave 8), and we are particularly dependent on information in this region to allow us to understand words. The second is that our hearing system not only does not have flat frequency response, but also (and this is truly weird) that the frequency response of our hearing system changes (that’s right: the bass and treble controls in our brain move!) as the loudness of the music changes, so that we listen with a constantly varying EQ system in our brain! Holy catfish!

	+enlarge
	Figure 12. The Frequency Response of our hearing at different sound levels (from Robinson and Dadson). These curves are approximately the inverse of the so-called “equal loudness contours” or Fletcher-Munson curves. The rolloff of the 40 dB SPL curve at 31.5 Hz. is approximately 38 dB, and off the scale of the graph.

The above chart reveals the following features about our hearing:

• hearing at all levels is fairly stable in the midrange;
• hearing is sharply boosted in Octaves 7 and 8, but by lesser amounts at lower levels;
• hearing response in Octaves 9 and 10 is wildly erratic, but generally rolled off;
• hearing response at low frequencies is steeply rolled off, and dramatically more so as the level falls off -- the perceived level at 31 Hz. at 100 dB SPL is only 82 dB SPL, while the same frequency at 40 dB SPL is perceived (not!) at the threshold of hearing, approximately 0 dB SPL!

Because our ability to hear bass improves as the Sound Pressure Level increases, small physical changes in bass level sound abnormally large to us, and it is really important to keep those levels consistent and comparatively loud in the mix. This requirement generally determines the level at which the song is finally mixed. Meanwhile, the peak at Octave 8 is nature’s way of saying, “Listen to the words, cat-breath.”* So don’t load up that range with lots of harmonics from other instruments.

There is one other big issue to keep in mind here. Our perception of loudness varies as a function of how much of the spectrum is covered. A sine wave with 1 milliWatt of power sounds considerably softer than 1 milliWatt of noise. If you wish to increase apparent loudness, increase the number of octaves of the spectrum you are using; if you want to decrease apparent loudness, decrease spectrum (particularly that area around 4 KHz.).