Footnote:
Clearly, one can invert this equation and obtain the azimuth from the ITD.
The auditory system must perform a more or less equivalent function in recovering
the azimuth from ITD information.
The accuracy with which this can be done depends on the circumstances.
For speech in normally reverberant rooms, typical human accuracies are on
the order of 10° to 20°. However, under optimum conditions, much
greater accuracy (on the order of 1°) is possible if the problem is
to decide merely whether or not a sound source moves. This is rather remarkable,
since it means that a change in arrival time of as little as 10 microseconds
is perceptible. (For comparison, the sampling rate for audio CD's is 44.1
kHz, which corresponds to a sampling interval of 22.7 microseconds. Thus,
in some circumstances, less than a one-sample delay is perceptible.)
It is also worth noting that if we use an interaural-polar
coordinate system and hold the azimuth constant, then we obtain a constant
value for the ITD. Thus, there is a simple one-to-one correspondence between
the ITD and the cone of constant azimuth, which is sometimes called the
"cone of confusion". This is not the case for
the vertical-polar system.
Finally, note that the ITD alone only constrains the source to be somewhere
on the cone of confusion. In particular, it is not sufficient for determining
whether the source is in front or in back.
Go up