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

Re: granular synthesis and auditory segmentation

Dear List,
  On Oct. 16, responding to my posting, Peter Cariani wrote:
>If I understand your processing correctly, you retain some constant
>proportion of successive half-waves, so that there is still a great deal
>of correlation structure between the half-waves that retains most of
>what is important to the auditory system.

   Yes.  The half-wave/granular way of looking at a signal waveform lets
you sort out and store interleaved granules from different sources and to
capture ten-microsecond time differences and rapid sequence transitions.
Granules avoid the limitations of the spectral window while carrying enough
information to sort them into meaningful categories.  Sources are
associated with their granules by identifying their categories and relative
times of arrival.  Correlation recovers the information in temporal
patterns of both tonal (periodic) and atonal (timbre) of each of the
intermixed time/space signal sources.

>When you have a receptor system that phase-locks to the stimulus, then
>the timings of discharges will more or less faithfully replicate the
>time structure of your half-waves, and the all-order intervals that
>are produced will more or less look like the autocorrelation of your

   Phase locking might be a way the ear correlates half-wave arrival times
with time-sequential events that carry meaningful patterns.

>In the correlational view, having all of those
>tuned filters makes the system much more robust
>in the face of background noise and multiple
>competing auditory objects.

   Here is an interesting point.  From the ear's point of view, is there a
solid definition of what constitutes background noise?  Is there a
difference between background noise and multiple competing auditory
objects?  If a noise-free environment is one that is devoid of acoustic
sources then we must be talking about, say, tinnitus.  If background noise
is composed of multiple auditory objects, then by what process does the ear
choose which background object is to be the "signal?"

>A full-blown autocorrelation-based theory of speech coding is possible
>that describes periodic and aperiodic speech sounds in terms of
>running autocorrelation structure. I think that this kind of
>theory would provide a way of thinking about some of the neural
>correlates of your experiments in granular synthesis.

   I agree, especially since what I've been doing uses a modified running
autocorrelator with ensemble-averaging referenced to the onset of each
arriving granule.


    John Bates

McGill is running a new version of LISTSERV (1.8d on Windows NT). 
Information is available on the WEB at http://www.mcgill.ca/cc/listserv