Version 1.0.0 of DRC 1.2.1 is now available.
This version is based on DRC 1.2 but differs in the way that it handles a particular class of grapheme-phoneme-correspondence (GPC) rules dubbed ‘split-grapheme’ rules. A split-grapheme rule is one in which the grapheme contains one or more gaps between its letters. The rule that produces the phoneme for the letters O and E in a word like LOBE is an example.
When such a rule is marked as a ‘middle-position’ rule it cannot match if the grapheme appears at the start of the word, but it can match if the grapheme runs to the end of the word, as it does in the word LOBE. Previous versions of DRC did not permit such a rule to match when it ran to the end of the word. Since such a rule does not produce the final phoneme, it makes sense that such rules be allowed to run to the end even when they are middle-position rules.
The English language data that accompanies this version has also been updated to version 1.1.7. The process used to build the vocabulary has been improved to eliminate certain inappropriate items, such as contractions. A handful of vocabulary items have been removed, or had their frequencies slightly lowered, as a result.
Builds are available for Mac OS X (170KB .tgz), Linux (507KB .tgz statically linked) and Windows (265KB .zip)