You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
45 lines
2.0 KiB
45 lines
2.0 KiB
Appendix C
|
|
|
|
Context-dependent rules for PHONET
|
|
|
|
This appendix presents the context-dependent rule set used in module PHONET.
|
|
|
|
C.1 Notation
|
|
The phonetic segment to parameter conversion rules are given in a form similar to
|
|
that of the lexical stress rules in Chapter 6. The following modifications are made
|
|
|
|
to the rule form described in Chapter 6:
|
|
e The symbol “S” is used to represent any phonetic segment. This
|
|
replaces the symbols “V” and “C” used in the previous set of rules.
|
|
e In addition to the features “+stress” and “-stress”, there is a set of fea-
|
|
tures used to classify phonetic segments according to general
|
|
|
|
properties. These features are listed in the next section.
|
|
The general form of a rule is as follows:
|
|
|
|
variable « value [ pattern
|
|
|
|
which means: “variable gets set to value in the context of pattern”. In addition to
|
|
the « operation, there are the T and | operations which mean to increase or
|
|
decrease (respectively) the value of variable by the amount value.
|
|
|
|
The variable is one of several parameters which hold state information about
|
|
the current phonetic segment. For example, “Target” is a table of target values for
|
|
each parameter at the end of the current segment. The rule:
|
|
|
|
Target[ave] 4 30/ [ +fricativi +voiced] [ -voswel]
|
|
|
|
means that the Target value for parameter avc gets decreased by 30dB, if the cur-
|
|
rent segment is a voiced fricative and the next segment is nonvocalic.
|
|
|
|
The overall structure of the program which implements these rules is as fol-
|
|
lows: The top level of the program is a loop which examines each phonetic seg-
|
|
ment from the input stream in sequence, one at a time. For each segment, the set
|
|
of state variables is initialized, then the rules are applied in sequence in the order
|
|
presented below. After the rules have been applied to the current segment, the in-
|
|
formation in the state variables is used to update the values of the output
|
|
parameters over the current time interval (from the start time of the current seg-
|
|
ment to the end time, as previously determined by PROSOD).
|
|
|
|
181
|