Azimuth Test using Spherical Model:
Good results in general. The negative angles were often misjudged to be
greater then what they really where.
|
Azimuth Test using HRTF
Very good results. The error is very small and the general trend of
the data is good. Judging from the results it is better then the
Spherical Model.
|
Elevation Test using HRTF
The results for this test are discouraging. Answers have a large
variance, and the mean hardly detects any perception of elevation.
Representation of elevation through an hrtf seems more difficult
than expected. Neverthless the results are definitely better than
what was found using the pinna model.
|
Space Azimuth Test using HRTF
Again results are very good. The Error is low for most values. Casual
inspection reviels that if reversals where condidered the results would
be good even for those points that currently seem incorrect. In addition
note that +180 and -180 are actually the same location. The data
although it may seem to be incorrect is in fact very accurate.
|
Space Elevation Test using HRTF
As before the results are poor. The error is very large and the
elevation is hardly percieved.
|
General Results
The detection of Azimuth angles appears to be achievable fairly
easily using either the HRTF or the spherical model. Elevation seems
much more difficult to simulate and elevation at extreme angles such
as -40 or +90. it is almost always interpreted incorrectly.
Front to back reversal are a big problem and have been thoroughly
examined in detail in lab environments that we could not recreate
given the scope and time contraints of this project. Additionally,
we would be hard pressed to find volunteers for these gruesome test.
These reversals may have accounted for some of the error encountered
in the Space Azimuth Test. In addition it should be noted that a spherical
model does not distinguish between front and back and hence cannot
be used to place a virtual sound source behind the listener.
|
System Comparison
It was difficult to find systems and implementations that we could
directly compare to our work. Most commercial systems are packaged with
other software (i.e. games, etc), and so a direct apples-to-apples
comparison was impossible. We found a system from
Wave Arts, Inc.
. The demo clips provided on their web site are very impressive. The
software package seems easy to use and allows many cool effects.
|
|