Babygruut

Added

- Support for Luxembourgish (credit: mbarnig)
- External pronunciation lexicons with <lexicon uri="...">

Fixed

- Choose first word pronunciation when all pronunciations have roles instead of last
- Word role is on <lexeme> instead of <grapheme>

Added

- Support for inline SSML pronunciation lexicons (<lexicon>, <lookup>)

Changed

- Moved English data files to separate Python package so core can be updated without large download
- With --ssml, input from stdin is assumed to be one document instead of lines (override with --stdin-format lines)

Fixed

- <sub> tag
- Avoid parsing non-finite numbers as words (nan, inf)

Added

- SSML support (see README for supported tags)
- Date, time, and initialism (e.g., T.T.S.) expansion
- Swahili (sw) language

Changed

- API and command-line interface (use gruut.sentences)

Removed

- Inline pronunciations, <number>_<format> (use SSML)

Fixed

- eSpeak lexicons and g2p models re-created without language flags

Added

- Arabic (ar) and Farsi (fa) languages
- Inline pronunciations with [[ phonemes ]] or {{ word s{eg}m{ents} }} (--inline)
- text2phonemes command for CLI (combined tokenize/phonemize)
- Support for Phonetisaurus grapheme-to-phoneme FSTs converted with fst2npz.py (--phonetisaurus)

Fixed

- bin scripts support speaker column in CSV
- Tests properly extend PYTHONPATH
- setup.py specifies README encoding
- Code formatting/linting includes bin scripts

Changed

- Require num2words fork for additional language support (https://github.com/rhasspy/num2words)
- Lexicons and CRF models added to git repo