Title:
|
Binaural hearing and lateralisation : the perception of interaural differences of amplitude and time
|
The perception of the direction of a sound source in the horizontal
plane is largely dependent upon the relative arrival times of salient
points in the waveform and upon the difference in amplitude at the two
ears. Other effects such as binaural release from masking are mediated
mainly by the percept of lateralisation.
In an extensive literature review the major experiments in binaural
unmasking. discrimination and lateralisation are introduced and the most
influential binaural detection and lateralisation models discussed. It
is argued that these models are all cross-correlation mechanisms
operating upon the differences between the firing patterns of the two
auditory nerves. A study of the response of the binaural system to
changes in its input would be a critical test of such models, so an
experiment to measure the threshold of a static tone in noise with
temporally varying Interaural phase was performed. The results suggest
that binaural processing is slow.
The extent of lateralisation of bandpass (10%) filtered clicks of
both low and high frequencies was studied with various interaural time
and amplitude differences. A novel feature of the research, apart from
the scaling technique used, was that subjects were encouraged to listen
for multiple images. These experiments are sensitive to the breakdown of
sensory fusion, and so pose a severe test for binaural models. Very
similar results at both low- (260 Hz) and high- (8000 Hz) frequencies
suggest a common lateralisation mechanism operating primarily upon
interaural onset time differences.
A binaural model is proposed which extends existing cross-correlation
models. Included is an auditory nerve model which adapts and
saturates. The other new element is coincidence detectors with significant
(1 ms) integration times, which more closely represent the temporal
Integration properties of real neural networks. The inclusion of the
auditory nerve model Is very succesful, but the extended coincidence
detectors prove less helpful. A single channel excitatory-inhibitory
model Is also discussed.
|