To compare the ability of subjects to locate sounds produced with
the modeled HRTFs (the Brown model) and the measured HRTFs (KEMAR dataset).
Experiment Setup
We have chosen a range of spatial locations to simulate using both the KEMAR dataset and the Brown model.
We selected these locations
because they provided a variety of positions to test azimuth cues, elevation cues, and a combination of the two.
We selected these locations at discrete intervals in azimuth and elevation making them far enough apart to be distinguishable.
We then
determined a random and consistent order to play these for our
subjects. We limited the spatial locations to the front half plane to correspond with the simplifying assumptions in developing the Brown model. Azimuth would therefore be limited to -90 to 90 and elevation limited to -90 to 90. The spatial locations that we selected to test were:
Azimuth
Elevation
Ordering Code
0
-40
a
0
0
b
0
40
c
0
80
d
30
0
e
-30
0
f
60
0
g
-60
0
h
90
0
i
-90
0
j
58
40
k
32
-40
l
-32
40
m
-58
-40
n
60
80
o
There were a total of thirty tests on 8 subjects.
For each location and each model we created a .wav file of pulses of
broadband noise.
Each pulse was Gaussian white noise bandlimited to 200-17000 Hz. Each
pulse
was 750 ms in duration and was enveloped at the beginning and end with a 20 ms sine squared curve.
The first two pulses of each sample were not filtered by the HRTF in
order to provide the listeners with a frame of reference. These
pulses should sound as if they are coming from inside the head.
The remaining four pulses were convolved with the
transfer function corresponding to the elevation and azimuth that
we were trying to test. The MATLAB code that generates our stimulus
can be found here
Experiment Procedure
Each person was given the same written instructions (provided here), and allowed to ask any questions.
Each person who was tested wore the same earphones and listened to the
files in the same order.
They were allowed to listen to each file no more than twice, and then
asked them to determine where they thought the sound came from.
We gave them the following options for azimuth: -90, -60, -30, 0, 30,
60, 90.
And the following options for elevation: -40, 0, 40, 80.