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

Re: Forming sequential objects

In response to Smoliar :

> Lerdahl and Jackendoff have problems

Don't we all. No theory of music or language that I know of is complete.
What is of interest experimentally is to test what they have formalized
at various levels of the theory. One level concerns melodic grouping,
though it is clear, and I agree with Smoliar on this point, other
levels of sectional grouping may be based on textural and figural
properties (see Clarke & Krumhansl in Music Perception, Deliege & El Ahmadi
in Psychology of Music). I think we will find that such things are
based on discontinuities in surface properties and are relatively
universal as segmentation procedures.

> edge detection in vision
This is probably the direct equivalent of durational and qualitative
discontinuity detection in audition. Where a discontinuity occurs, we
establish a boundary, though where that boundary falls exactly may be
ambiguous in melodic grouping as Deliege has shown. The edge of a
texture may also be fuzzy depending on the temporal properties of the
musical texture and on the degree of similarity of two adjoining
textures. If I recall correctly, John MacKay did some work on this in
the realm of contemporary music at U.C. San Diego in the early 80s.

> the interesting research question is one of whether or not existing computer
> vision techniques can be adapted to audio signals.
I think we'd better figure out more systematically what features human
listeners are sensitive to before we run off and start modeling the beast.

Stephen McAdams