The phonetician's assistant
- First public release of the Perkins source code. Now under the GNU GPL v3 License.
1.0.1 to 1.0.3
- Various unpublished internal changes.
- Multiple identical unaccented vowels (e.g. "aa", "aaa", "aaaa" are now reduced to a single vowel. Doing this with two vowels is useful for the odd Spanish word and surname (e.g. "zoológico", "Saa"), which only newscasters who strive to sound like TTS systems actually pronounce with two vowels, as well as for the occasional foreign term or name ("Baader", "Groot"). Reducing three or more indentical unaccented vowels to just one is a handy way of dealing with the seemingly limitless variety and length of interjections that are out there in the wild (e.g. "jajajaaaa", "jajajaaaaaaaaaa", "jajajaaaaaaaaaaaaaaaaaaaaa" and so on).
- Added the -och and -oye options, which allow you to force the use of the single character /ʧ/ and /ʤ/ symbols no matter what other options are chosen. Previously, these were linked to the -nomc (no multiple characters) option, which would not let you have /ʧ/ plus dentals with diacritics. That was not a good design choice, and this lets you get around it.
- Added a series of additional lexical exceptions to deal with specific items such as "sd", "xd", "tour", etc.
- Updated the online help to reflect the above options, as well as to highlight the -nosd (no syllable dots) option, which was not immediately apparent, as it's actually the negation of another command, -sd.
- Fixed a long-standing bug that was probably implemented as a way to properly process incorrectly written but unambiguous words, such as "samurai" (which should actually be transcribed as /sa.'mu.ɾaj/), which should actually be written as "samurái"). This affected a vanishingly small number of non-Chilean Spanish words, but it laid waste to many Chilean voseo conjugations, namely those that end in orthographic "i" (e.g. "estabai", "hubierai").