Generated by python scripts/benchmark.py from the committed gold
fixtures (tests/data/gold/), which are fixed-seed samples of
TigreGotico/portuguese-unified-pronunciation-lexicon.
Rules-only: the lexicon exception path is emptied, so the number
measures grapheme-rule quality, not lexicon lookup coverage — see
benchmarking.md.
| Dialect | Gold region | Words | PER | Word accuracy |
|---|---|---|---|---|
| pt-AO | pt-AO | 1000 | 0.2603 | 0.0660 |
| pt-BR | pt-BR | 1000 | 0.2523 | 0.0770 |
| pt-BR-x-rio-janeiro | pt-BR-x-riodejaneiro | 1000 | 0.1653 | 0.1790 |
| pt-BR-x-sao-paulo | pt-BR-x-saopaulo | 1000 | 0.1683 | 0.1820 |
| pt-MZ | pt-MZ-x-maputo | 1000 | 0.2360 | 0.1540 |
| pt-PT | pt-PT | 1000 | 0.2294 | 0.1540 |
| pt-PT-x-lisbon | pt-PT-x-lisboa | 1000 | 0.1007 | 0.4330 |
| pt-TL | pt-TL-x-dili | 1000 | 0.2488 | 0.1070 |