thai_aksharamukha.yml 1.2 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849
  1. ---
  2. general:
  3. name: Thai (Aksharamukha)
  4. comment: "Note: not ALA-LC compliant."
  5. case_sensitive: false
  6. options:
  7. - id: ThaiTranscription
  8. label: Thai Orthography
  9. description: พุทฺธ → พุทธะ
  10. type: boolean
  11. default: false
  12. - id: ThaiSajjhayaOrthography
  13. label: Sajjhāya orthography
  14. description: พุทฺธ → พุท์ธ
  15. type: boolean
  16. default: false
  17. - id: ThaiSajjhayawithA
  18. label: Nativized sajjhaya
  19. description: พุทฺธํ → พุท์ธัง
  20. type: boolean
  21. default: false
  22. - id: ThaiNativeConsonants
  23. label: Thai phonetic
  24. description: พุทฺธตฺว → บุดธะต͜วะ
  25. type: boolean
  26. default: false
  27. - id: ThaiVisargaSaraA
  28. label: Sara a ะ as Visarga
  29. description: พุทฺธ → พุทธะ
  30. type: boolean
  31. default: false
  32. script_to_roman:
  33. hooks:
  34. post_config:
  35. -
  36. - asian_tokenizer.s2r_tokenize
  37. - model: "KoichiYasuoka/roberta-base-thai-spm-upos"
  38. -
  39. - aksharamukha.romanizer.s2r_post_config
  40. - src_script: "Thai"
  41. roman_to_script:
  42. hooks:
  43. post_config:
  44. -
  45. - aksharamukha.romanizer.r2s_post_config
  46. - dest_script: "Thai"