- sorry for the silliness of my previous post-- I
got a little overexcited.
I would really appreciate the advice of anyone on how
to improve the performance of a pause detection script--
and also to query the number of intervals now that
the Get number of intervals... command no longer seeems to work
on the objects in Mietta's pause script..
Also, would there be any value in a high level functional
performance grammar with constraints like
* Phonogical Error (an integration of all faithfulness
(A constraint which gives rise to the dynamics of pitch
with the noise process cognitive load (and no doubt very
complex in structure)
.-- such a
simple grammar could surely be learnt with small amounts
of data and thus could characterise individual speakers from
I was wondering if
anyone has any idea of the bounds for a senible size training
set to train a grammar with only four constraints and a few
categories? - or how to bootstrap estimates of harmony
when the corpus is limited?