138 Audition
M 0 S 0. If the masker is M 0 , but the signal is presented with a
180
( radians) interaural phase difference, the condition is
M 0 S. The threshold for detection of the signal in the M 0 S
condition is 15–18 dB lower than it is in the M 0 S 0 condition
(the MLD is 15–18 dB). The MLD has been studied for a
wide variety of stimulus conditions and interaural configura-
tions, and the MLD is always positive when the signal and
masker have a different set of interaural differences, as com-
pared to conditions in which the signal and masker have the
same set of interaural differences.
Because an interaural difference is associated with a hori-
zontal location in space, signals and maskers that have differ-
ent interaural differences are similar to stimuli that are at
different positions in space. Thus, the results from the MLD
literature suggest that when the signal and masker are in dif-
ferent spatial locations (have different interaural differences),
signal threshold is lower than when the signal and masker oc-
cupy the same spatial locations (have the same interaural dif-
ferences). Such threshold differences do exist when signals
and maskers are presented from loudspeakers in real-world
spaces (Gilkey & Andersen, 1997). These results appear con-
sistent with Cherry’s observation about the role spatial sepa-
ration plays in solving the cocktail party problem. Models
that are variations of the coincidence models used to account
for processing interaural time differences have also been suc-
cessful in accounting for a great deal of the data from the
MLD literature (Colburn & Durlach, 1978).
Pitch and Timbre: Harmonicity and
Temporal Regularity
Pitch is that subjective attribute of sound that varies along a
low-high dimension and is highly correlated with the spectral
content of sound. The pitch of a target sound is often given in
terms of hertz, such that the pitch of a target sound is xHz, if
a tone of xHz is judged perceptually equal in pitch to the tar-
get sound. Musical scales, such as the 12-note scale, can also
be used to denote the pitch of a sound.
Timbre is defined as that subjective attribute of a sound
that differentiates two sounds that are otherwise equal in
pitch, loudness, and duration. Thus, the difference between
the sound from a cello playing the note G for the same dura-
tion and loudness as the sound from a violin playing the same
note G, is said to be a difference in timbre. The sound of the
cello differs in timbre from that of a violin. There are no units
for measuring timbre, and timbre is often correlated with the
spectral or temporal complexity of the sound.
Although the pitch of a sound is often highly correlated
with frequencies that are the most intense in a sound’s spec-
trum, many complex sounds produce a strong pitch in the
absence of such a concentration of spectral energy. Consider a
complex sound with frequency components of 300, 400, 500,
and 600 Hz. This sound will often have a 100-Hz pitch, even
though there is no spectral component at 100 Hz. Note that
100 Hz is the fundamental of this sound (all of the existing
spectral components are harmonics of a 100-Hz fundamental),
but the fundamental is missing. Thus, this type of complex
pitch is referred to as the “pitch of the missing fundamental.”
Many sound sources (e.g., most musical instruments) contain
a spectrum of harmonics. The pitches associated with these
sounds are derivatives of the pitch of the missing fundamental.
The stimulus described above that leads to the pitch of the
missing fundamental will often have a periodic time envelope,
which in this case will have a 100-Hz repetition (a 10-ms
period). Thus, the pitch may be associated with the temporal
regularity in the envelope. However, stimuli with very little
envelope periodicity can still produce a complex pitch like
that of the pitch of the missing fundamental. Such stimuli may
not have a smooth spectrum like that of the tonal complex
described above. Thus, neither envelope periodicity nor a
smooth spectrum appear to be necessary and sufficient condi-
tions for producing a complex pitch. However, such stimuli
without periodic temporal envelopes may contain a tempo-
rally regular, but nonperiodic, fine structure that may be the
basis for complex pitch (an analysis of this stimulus, such
as autocorrelation, will reveal this otherwise difficult-to-
determine temporal regularity; see Yost, 1996).
In addition to influencing the pitch of complex sounds, har-
monic structure also influences timbre. Thus, a complex
harmonic sound with high-amplitude, high-frequency har-
monics may have a brighter timbre than a complex sound
with high-amplitude, low-frequency harmonics, which would
have a dull timbre. Certain forms of temporal regularity (e.g.,
noise vs. periodic sounds) can also influence a sound’s timbre.
Therefore, harmonic structure and temporal regularity are
important stimulus properties that help determine the pitch
and timbre of complex sounds. Complex sounds differ in
pitch and timbre, and, as such, these two subjective attributes
may allow for sound source segregation. Indeed, complex
pitch and timbre have both been used to segregate sound
sources in auditory stream experiments (Bregman, 1990).
The two-vowel paradigm (Summerfield & Assmann, 1991;
Yost & Sheft, 1993) is another procedure used to study the in-
fluence of harmonicity on sound source segregation. In the
two-vowel procedure, two artificially generated (via com-
puter) vowels are mixed. Often it is difficult or impossible
to identify the two vowels generated in this manner. Any stim-
ulus manipulation that allows for vowel recognition in the
two-vowel stimulus is arguably a crucial stimulus condition
for sound source segregation. If the fundamental voicing