What's Changed
English ITN
- feat: add English ITN core rules (cardinal, ordinal, decimal) (#355) by @pengzhendong
- feat: add English ITN extended rules (time, whitelist) (#356) by @pengzhendong
- feat: English ITN with full rule coverage — 469/470 NeMo tests pass (#358) by @pengzhendong
- feat: English ITN enhancements from NeMo reference (#361) by @pengzhendong
- Time: numeric "past", "till" alias, o'clock variants
- Telephone: "triple X" support
- Date: BCE/CE, H1/H2, century/millennium ranges
- Measure: finite pluralization, irregular plurals
Chinese TN/ITN
- feat: add TrainNumber rule for Chinese ITN (#335) by @pengzhendong
- feat: support "N个X" repeat digit pattern in Chinese ITN (#336) by @pengzhendong
- feat: add 亿级 support and strip commas for Chinese TN (#337) by @pengzhendong
- feat: support nbest output in normalize/tag (#348) by @pengzhendong
- add remove_interjections option for chinese itn (#324) by @duj12
English TN
- feat: add Serial rule for English TN (#354) by @pengzhendong
Bug Fixes
- fix: resolve 16 failing unit tests (#332) by @pengzhendong
- fix: restore Chinese ITN measure fields in TokenParser (#357) by @pengzhendong
- fix: OpenFst compatibility + add C++ runtime CI (#360) by @pengzhendong
- fix: support L (liter) unit and number+unit denominator in Measure (#338) by @pengzhendong
- fix: remove contraction expansions from whitelist (#339) by @pengzhendong
- fix: add missing space in comma-separated numbers (#342) by @pengzhendong
- fix: disambiguate "X点零Y" between time and decimal (#343) by @pengzhendong
- fix: convert digit after 块 regardless of enable_0_to_9 (#344) by @pengzhendong
- fix: don't convert - to 减 in non-math context (#347) by @pengzhendong
- fix: only use digit-by-digit reading for 4-digit years (#349) by @pengzhendong
- fix: enumerate known short phone numbers (#351) by @pengzhendong
- fix: remove 10000 from phone.tsv (#353) by @pengzhendong
- fix: Resolve installation bug #292 (#325) by @WhizZest
- fix: relax pynini version constraint to >=2.1.6 (#345) by @pengzhendong
- fix: add Japanese data to package-data (#346) by @pengzhendong
Refactoring & CI
- refactor: share rule instances to eliminate redundant FST construction (#333) by @pengzhendong
- refactor: code review cleanup and bug fixes (#350) by @pengzhendong
- refactor: fix encoding and move str2bool to shared utils (#352) by @pengzhendong
- ci: trigger unit tests on tsv file changes (#341) by @pengzhendong
- update wheels.yml to actions/upload-artifact@v4 (#326) by @freddy-z
- Update runtime deps & CXX standard (#331) by @mallocnew
New Contributors
- @duj12 made their first contribution in #324
- @WhizZest made their first contribution in #325
- @freddy-z made their first contribution in #326
- @mallocnew made their first contribution in #331
Full Changelog: v1.0.5...v1.2.0
Contributors
pengzhendong, duj12, and 3 other contributors