Following our phone conversation, here is a list with the changes in the
xml-files Cambridge would like to have in order to use PROCSY as a
copysynthesizer:
++ Release should be marked for affricates as well as stops.
++ Fricatives and plosives should stop at main label (e.g. f or p ) and
not take any additional alternative boundary labels into account (i.e.
labels like f_1 or f_2 should be ignored).
Devoicing labels (e.g. z_0) might come after the main label but should be
part of the fricative, of course. The easiest way to achieve this change
seems to me simply to ignore all labels with a number bigger than 0 for
all fricatives and plosives in the alignment process ( e.g. wih awk '$1
!~ /_1/ || $1 !~ /[fvszSZtd..]/)
++ Laryngalizations should be ignored as well, i.e. filter out V_L
(vowelsymbol_laryngealization) labels before running the alignment.
We hope we've identified everything, so it can all be done at once. Could
you give us an idea when the revised xml-files would be ready?
Sebastian