Since Jill didn't get the attachment (and I couldn't read it they way
it it's in the archive) I assume something went wrong, so I give it a
second go (This time in plain ascii to make things easy).
***********************************************************************
Notes on labelling
4-9-98 Sebastian, Sarah, Rachel
Examples can be seen in the label files which we will put on the
ProSynth ftp site, as soon as we have them and found out how to do that.
We'll let you know when we've done it.
a) rename labels:
t_burst -> tc (end of t closure)
for bursts without preceding closure, i.e. a transient see (l) below.
t remains as end of aspiration and therefore end of t.
b) new labels:
general: label each significant acoustic segment. Precede each sub-label
by main phoneme label "lab" and "_". Thus the label name after "_"
describes attributes that aren't part of the canonical phonemic
description, e.g. voicing in voiceless stop closure, or parts of
a phone with multiple acoustic segments (e.g. a voiced stop).
(i) t_v end of voiced part (also ?_v, d_v, k_v, g_v, D_v etc.)
(ii) Transients
-- bursts preceded by silence (closure): lab_c[_#] where
optional # indicates multiple releases, each preceded by silence
(# is a number: normally number in ascending order from left)
except that
the main burst is lab_c (no number) and is the one where the
continuity with next segment starts (noise, aspiration, ringing,
whatever).
So t_c_# can only precede t_c.
-- transients that are not preceded by silence: lab_b[_#]
found e.g. in D, and in stops after the main burst (lab_c)
(iii) _0 denotes voiceless part of a phonemically voiced segment
(e.g D_0, m_0, a_0)
(iv) if we find extensive ambiguous regions lab_n is used
s_1 maximum periodicity
s_2 minimum periodicity
lab lab_n lab_(n+1) lab is most preferred, the other ones are less
preferred, so the n suggests position in a hierarchy of
preference
(v) lab_! means something interesting is happining here which is way too
complex to label properly. To be used as rarely as possible.
c) special treatment for r
r means end of reasonable stable central region of the r
suggestion (that we abandoned as too complex):
r_s_fx -> start of transition of formant x into the r
r_e_fx -> end of transition of formant x out of the r
details to be added by York?!
only interesting end of transitions will be mentioned
c) Consequences of new labels:
(i) D is divided as appropriate into:
D (voiced and friction),
D_v (voiced but closed, no friction),
D_c (no friction no voice),
D_0 (unvoiced with friction)
and
_b is added to any non-silent segment that ends with a burst,
e.g. D_b, D_0_b, D_v_b, etc.
T follows the same logic.
(ii)
prosynth slogan 3b.1: "label from the left"
e.g. overlapping voice and frication
in a V-F-Sequence call it V V_1 ...
in a F-V-Sequence call it F F_1 ...
(iii)
Pause mark at beginning of sentence can be coincident with another label.
So, if utterance starts with a burst we would call that X_c indicating
that there is an end of closure. On the other hand it is still the
end of the pause.
******************************************************************
eoNoL
Sorry for the problems,
Sebastian
*********************************************************************
Sebastian Heid Email: sh276@cam.ac.uk
Phonetics Laboratory Phone: +44 1223 33 50 50
Dept. of Linguistics Fax: +44 1223 33 50 53
University of Cambridge
Sidgwick Avenue
Cambridge CB3 9DA
United Kingdom
*********************************************************************