[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

David Huron's calculations

Here are two comments on David Huron's interesting description of a
segregation model, dated 6 December.

First, there is a danger in moving so quickly to a place-based model for
spectral and virtual pitch because the model ignores the timing information
that can signal synchrony or asynchrony amongst the spectral components.
The problem is magnified if one uses a Terhardt-type model where spectral
pitches are shifted up and down the tonotopic axis by peripheral
interactions. After a signal has been put through that kind of front end,
the timing information is lost forever.

The conceptual advantage of retaining the timing information is that the
asynchrony associated with inharmonic partials can, in principle, be
treated by the same processer that pays attention to the synchrony of
onsets and modulation. We use a model like that to study the (admittedly
very simple) mistuned harmonic experiment.

Second, it certainly does seem right to begin with pitch as the basic
principle whereby individual harmonics are integrated into single entities.
Pitch is the "great organizer." But, in order to do that, one needs a
self-consistent solution. If a virtual pitch is calculated from component
frequencies (or from spectral pitches) in a way that is independent of the
decision concerning the integration/segregation of the components, then
pitch is not really accorded its organizing role.