Sound localization

What are the factors that

determine how well we can tell
where a sound is coming from?
 Bottom line

Acoustics, peripheral coding, and

central processing are all
important in sound localization.
Importance of sound localization
• Clear survival value: evolutionary
importance
• Involves comparisons between ears and
across the spectrum: requires a brain
 Sound localization: the important
phenomena
• Effects of position in azimuth
• Effects of frequency
• Effects of position in elevation
The minimum audible angle
(MAA)
Sound 1
Sound 2

Speakers

Which sound came

from the right?

Threshold for detecting a change

in spatial location of a sound
source
 MAA at different positions

From Gelfand (1998)

MAA in azimuth

Judged position

MAA
 MAAazimuth as a function of
frequency

From Blauert (1983)

 MAA in elevation

From Blauert (1983)

 The MAA is
• Better at midline than to the sides in
azimuth
• Not so good for sounds around 1500 Hz
• Not as good in elevation as in azimuth

WHY?
Explanations for the characteristics
of the MAA

• Acoustic cues
available
• Peripheral coding
• Central processing
 Acoustic cues used in
localization
• Interaural differences
– Interaural intensity differences (IIDs)
– Interaural time differences (ITDs)
• Spectral shape cues
 Interaural intensity differences

From Gelfand (1998)

Interaural time differences
Why are we better at “straight
ahead” than off to the sides?
 Cone of confusion

From Gelfand (1998)

 Cone of confusion

150 degrees

30 degrees

30 degrees

150 degrees
 Why are we better at “straight
ahead” than off to the sides?

• Cone of confusion, front-back

confusions
• In the brain, more neurons respond to
“straight ahead” than to “off-to-the side”.
The brain can afford to devote
more neurons to localizing straight
ahead because

• we can turn our heads

• most sounds come from straight ahead
• it isn’t necessary to localize sounds
accurately off to the side
• none of the above; it can’t afford to do this
 IIDs and frequency

From Gelfand (1998)

 Why IIDs depend on frequency

1600 Hz

From Gelfand (1998)

 The reason that sound is more
intense at the ear close to the
sound source is
• the inverse square law; sound has to
travel farther to the far ear
• it takes longer for sound to travel farther
• the head absorbs and refects sound on
the side closer to the source
• the head absorbs and refects sound on
the side farther from the source
If there is a talker on your left and a
talker on your right, and you really
want to hear the talker on your
right, you should
• turn your nose toward the talker on the
right
• turn your nose toward the talker on the left
• turn your left ear to the left talker and your
right ear to the right talker (and listen to
your right ear)
 Interaural time differences

From Gelfand (1998)

 ITDs and frequency: How
could your brain know if a
sound arrived at one ear .2 ms
later than at the other?
 Phase ambiguity
1000 Hz: phase difference is 216 degrees

4000 Hz: phase difference is 864 degrees but looks like 144 deg

.6 ms

.1 ms
 Phase ambiguity
When the interaural delay is
longer than the period of the tone,
then interaural time comparison
gives an ambiguous result.

The comparison gives a result of x degrees phase

difference, but the real difference could be 360+x or
720+x, etc.
 Interaural onset time
comparisons

.6 ms

.6 ms

How about the frst (onset) response?

It can be used but not all sounds have abrupt onsets
and it only happens once.
 Interaural time differences
Maximum = .65 ms

F = 1/p

F = 1/.65

F = 1538 Hz

(and any higher frequency

From Gelfand (1998)

 The reason that phase ambiguity
occurs is
• the ear does not encode the starting
phase
• the ear provides no information about
phase
• phase locking does not occur above 5000
Hz
• people use the place code for frequecies
above 2000 Hz
 Interaural cues and frequency

IID
Cue goodness

ITD

250 500 1000 2000 4000 8000 16000

Neither cue is so good around

1500 Hz.
How do we know that (other)
characteristics of the auditory
system don’t also contribute?
 Lateralization experiments

Interaural differences under earphones create the perception o

sound source located inside the head, at a position determined
the interaural difference.
Interaural intensity difference
discrimination
Good performance
15 dB

9 dB

0 dB

From Gelfand (1998)

We know it’s the acoustics in the
case of IIDs, because if we
artifcially create IIDs at low
frequencies, people hear sound
source at different locations.
 Interaural time (phase) difference
discrimination

Still can’t do high

frequency

Better at small
phase separations
(straight ahead)

From Gelfand (1998)

 We know it’s the auditory system
in the case of ITDs, because if we
“artifcially” create ITDs at high
frequencies, people still can’t tell
what the sound source location is.
 AM lateralization

From Yost (1994)

We know there is a contribution of
the auditory system in the case of
differences between positions,
because if we artifcially create
different positions, people can still
“lateralize” better for midline than
for lateral positions.
Explanations for the characteristics
of the MAA

• Acoustic cues
available
• Peripheral coding
• Central processing
 The MAA is
• Better at midline than to the sides in
azimuth
• Not so good for sounds around 1500 Hz
• Not as good in elevation as in azimuth

WHY?
What are the acoustic cues to
sound elevation?

Sound source moving along

an arc directly overhead at
midline

No interaural differences
 What are the acoustic cues to
sound elevation?
Elevation - no
Azimuth - yes, but front-back
confusions
Elevation - no
Azimuth - yes
Elevation - yes
Azimuth - yes

Localization in elevation requires

pinnas
From Blauert (1983)
 What do pinnas do for us?

The acoustic cue used to localize in elevation is spectral shape.

From Gelfand (1998)
That localization is less precise in
elevation than in azimuth
suggests that spectral shape is
not as good a cue to location as
interaural differences.
 Localization in azimuth with
one ear is similar in precision
to localization in elevation.

From Gelfand (1998)

Spectral shape cues are available, but used as supplemental

Information for localization in azimuth.
Another role of spectral shape
cues

Unprocessed sound Sound shaped

by “HRTF flters”
Reason that localization in azimuth
is better straight ahead than off to
the side
• acoustic cues available
• peripheral coding
• central processing
Reason that localization in azimuth
is not so good for sounds around
1500 Hz
• acoustic cues available
• peripheral coding
• central processing
Reason that localization in
elevation is not as good as in
azimuth
• acoustic cues available
• peripheral coding
• central processing
 Conclusions
• The cues to sound location are
interaural time and intensity differences
and spectral shape.
• Interaural intensity cues are primarily
used at high frequencies, due to
acoustic limitations.
• Interaural time cues are primarily used
at low frequencies, due to limitations in
peripheral coding of sound, although
they can be used to localize amplitude
modulated sounds.
 Conclusions (continued)
• Spectral cues are used in localization in
elevation, to resolve front-back confusions,
and to produce the perception of a sound
in space.
• Neural processing limitations make us
more sensitive in sound localization
straight ahead.
 Text sources
• Blauert, (1983). Spatial hearing: The psychophysics of
human sound localization. Cambridge, MA: MIT Press.
• Gelfand, S.A. (1998) Hearing: An introduction to
psychological and physiological acoustics. New York:
Marcel Dekker.
• Yost, W.A. (1994) Fundamentals of hearing: an
introduction. San Diego: Academic Press.