Page 85 - Master Handbook of Acoustics

that of long pulses or steady tones or noise. This 100 msec appears to be the maximum integrating
time or the time constant of the human ear. In particular, events occurring within 35 msec, such as
reflections from walls, are integrated by the ear with regard to level. This shows that the ear responds
to sound energy averaged over time.

Figure 4-12 shows that our ears are less sensitive to short transients such as peaks in sound level.
This has a direct bearing on understanding speech. The consonants of speech determine the meaning
of many words. For instance, the only differences between bat, bad, back, bass, ban, and bath are the
consonants at the end. The words led, red, shed, bed, fed, and wed have the all-important consonants
at the beginning. No matter where they occur, consonants are transients having durations on the order
of 5 to 15 msec. Figure 4-12 shows that transients this short must be louder to be comparable to

longer sounds. In the above words, each consonant is not only much shorter than the rest of the word;
it is also at a lower level. Thus we need good listening conditions to distinguish between such sets of
words. Too much background noise or too much reverberation can seriously impair the intelligibility
of speech because they can mask important, lower-level consonants.

Audibility of Loudness Changes

As we have seen, the ear is sensitive to a wide dynamic range of sounds from the softest to loudest.
Within that range, the ear is relatively sensitive to small changes in loudness. For example, steps of 5
dB are definitely audible, whereas steps of 0.5 dB may be inaudible, depending on circumstances;
detecting differences in intensity varies somewhat with frequency and also with sound level. For

example, at 1 kHz, for very low levels, a 3-dB change is the least change detectable by the ear, but at
high levels the ear can detect a 0.25-dB change. (The former statistic is why many recording
engineers mix at loud playback volumes.) As another example, a very low-level, 35-Hz tone requires
a 9-dB level change to be detectable. For the important mid-frequency range and for commonly used
levels, the minimum detectable change in level that the ear can detect is about 2 dB. In most cases, at
least in acoustical design, making level changes in increments less than these is usually unnecessary.

Pitch versus Frequency

Pitch is a subjective term. It is chiefly a function of frequency, but it is not linearly related to it.
Because pitch is somewhat different from frequency, it requires another subjective unit, the mel.
Frequency is a physical term measured in hertz. Although a soft 1-kHz signal is still 1 kHz if you

increase its level, the pitch of a sound may depend on sound-pressure level. A reference pitch of
1,000 mels is defined as the pitch of a 1-kHz tone with a sound-pressure level of 60 dB. The
relationship between pitch and frequency, determined by experiments with juries of listeners, is
shown in Fig. 4-13. On the experimental curve, 1,000 mels coincides with 1 kHz; thus the sound-
pressure level for this curve is 60 dB. The shape of the curve of Fig. 4-13 is similar to a plot of

position along the basilar membrane as a function of frequency. This suggests that pitch is related to
action on this membrane.