Binaural masking of amplitude modulation

A new concept concerned with the transformation of acoustic
stimuli in the auditory system postulates the existence of a form of spectral
analysis applied to the amplitude changes of the stimuli. It is assumed that
this analysis takes place in the so-called modulation filters, i.e. bandpass
linear filters tuned to different rates of the amplitude changes. The most
striking argument supporting this idea is an effect of masking in the amplitude
modulation domain whose nature can be easily explained basing on this concept.
As the modulation filters are situated on the higher levels of the auditory
system, it is also assumed that this form of masking is entirely a central
process. However, most of the studies concerned with masking in the modulation
domain used monaural listening only. Therefore, the main purpose of the
presented here experiments was to investigate whether this type of masking is
entirely a central process. Using a Three-Alternative Forced-Choice (3AFC)
procedure the binaural and monaural masked thresholds of amplitude modulation
were determined. A sinusoidal carrier at a frequency of 4 kHz was amplitude
modulated by a specially designed band of noise characterized by a very low
value of the crest factor, which was used as a masking signals. Different
bandwidths of the modulating masking signals were used as well as different
center frequencies to investigate whether the masking patterns in the modulation
domain depend on the masker bandwidth and its center frequency. The modulating
target (masked) signal was a pure tone at a frequency range from 2 to 256 Hz.
Both modulating signals were applied to the same sinusoidal carrier signal. The
most effective masking was noticed when the rate of the sinusoidal modulation
was close to the center frequency of the masking signal or when it was in its
spectral range and decreased outside of this range. The character of this
dependence confirms the existence of some form of a frequency selectivity in the
modulation rate domain similarly to the audible frequency domain. The thresholds
for monaural and binaural listening were very close to each other. This implies
that masking in the modulation domain is a central process.