Digital Sound & Music: Concepts, Applications, & Science, Chapter 5, last updated 6/25/2013
184.108.40.206 Psychoacoustics and Perceptual Encoding
Psychoacoustics is the study of how the human ears and brain perceive sound. Psychoacoustical
experiments have shown that human hearing is nonlinear in a number of ways, including
perception of octaves, perception of loudness, and frequency resolution.
For a note C2 that is one octave higher than a note C1, the frequency of C2 is twice the
frequency of C1. This implies that the frequency width of a lower octave is smaller than the
frequency width of a higher octave. For example, the octave C5 to C6 is twice the width of C4
to C5. However, the octaves are perceived as being the same width in human hearing. In this
sense, our perception of octaves is nonlinear.
Humans hear best (i.e., have the most sensitivity to amplitude) in the range of about 1000
to 5000 Hz, which is close to the range of the human voice. We hear less well at both ends of the
frequency spectrum. This is illustrated in Figure 5.45, which shows the shape of the threshold of
hearing plotted across frequencies.
Figure 5.45 Threshold of hearing plotted over frequencies in range of human hearing
Human ability to distinguish between frequencies decreases nonlinearly from low to high
frequencies. At the very lowest audible frequencies, we can tell the difference between pitches
that are only a few Hz apart, while at high frequencies the pitches must be separated by more
than 100 Hz before we notice a difference. This difference in frequency sensitivity arises from
the fact that the inner ear is divided into critical bands. Each band is tuned to a range of
frequencies. Critical bands for low frequencies are narrower than those for high ones. Between
frequencies of 1 and 500 Hz, critical bands have a width of about 100 Hz or less, whereas the
width of the critical band at the highest audible frequency is about 4000 Hz.
The goal of applying psychoacoustics to compression methods is to determine the
components of sounds that human ears don’t perceive very well, if at all. These are the parts that
can be discarded, thereby decreasing the amount of data that must be stored in digitized sound.
An understanding of critical bands within the human ear helps in this regard. Within a band, a