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.
44 lines
2.0 KiB
44 lines
2.0 KiB
6
|
|
|
|
Letter-to-sound and lexical stress
|
|
|
|
6.1 Overview
|
|
|
|
In order to convert unrestricted text to speech, it is necessary to have a scheme
|
|
which stipulates a pronunciation for words not analyzable by the lexical analysis
|
|
algorithm. This comprehensiveness is provided by the letter-to-sound section of
|
|
SOUNDI1. The letter strings which it receives are converted into stressed phonetic
|
|
segment label strings (hereafter referred to as segment strings) using two sets of
|
|
ordered phonological rules (Hunnicutt, 1976b). The first set to be applied converts
|
|
letters to phonetic segments, first stripping affixes, then converting consonants,
|
|
and finally converting vowels and affixes. The second set applies an ordered set of
|
|
rules which determine the stress contour of the segment string.
|
|
|
|
These rules were developed by a process of extensive statistical analysis of
|
|
English words. The form of the rules reflects the fact that pronunciation of vowels
|
|
and vowel digraphs, consonants and consonant clusters, and prefixes and suffixes
|
|
is highly dependent upon context. The method of ordering rules allows converted
|
|
strings which are highly dependable to be used as context for those requiring a
|
|
more complex framework. Detailed studies of allowable suffix combinations, and
|
|
the effect of suffixation on stress and vowel quality, have also provided for more
|
|
reliable results.
|
|
|
|
This component is integral to SOUNDI1 described in the previous chapter and
|
|
processes words which were not segmented by DECOMP. Input and output for-
|
|
mats are described in that chapter.
|
|
|
|
6.2 Letter-to-sound
|
|
|
|
6.2.1 Operation
|
|
|
|
The conversion of a letter string to a phonetic segment string in the letter-to-sound
|
|
program proceeds in three stages. In the first stage, prefixes and suffixes are
|
|
detected. Such affixes appear in the list of phonological rules. Each is classified
|
|
|
|
according to:
|
|
1. its possible parts of speech,
|
|
2. the possible parts of speech of a suffix preceding it,
|
|
3. its restriction or lack of restriction to word-final position, and
|
|
|
|
57
|