fraction.py 2.3 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758
  1. import pynini
  2. from fun_text_processing.text_normalization.en.graph_utils import (
  3. DAMO_NOT_QUOTE,
  4. DAMO_SIGMA,
  5. GraphFst,
  6. delete_preserve_order,
  7. insert_space,
  8. )
  9. from pynini.lib import pynutil
  10. class FractionFst(GraphFst):
  11. """
  12. Finite state transducer for verbalizing fraction
  13. e.g. fraction { integer: "drei" numerator: "eins" denominator: "zwei" }-> drei ein halb
  14. e.g. fraction { numerator: "vier" denominator: "zwei" } -> vier halbe
  15. e.g. fraction { numerator: "drei" denominator: "vier" } -> drei viertel
  16. Args:
  17. ordinal: ordinal GraphFst
  18. deterministic: if True will provide a single transduction option,
  19. for False multiple transduction are generated (used for audio-based normalization)
  20. """
  21. def __init__(self, ordinal: GraphFst, deterministic: bool = True):
  22. super().__init__(name="fraction", kind="verbalize", deterministic=deterministic)
  23. optional_sign = pynini.closure(pynini.cross("negative: \"true\"", "minus ") + pynutil.delete(" "), 0, 1)
  24. change_one = pynini.cdrewrite(
  25. pynutil.add_weight(pynini.cross("eins", "ein"), weight=-0.0001), "[BOS]", "[EOS]", DAMO_SIGMA
  26. )
  27. change_numerator_two = pynini.cdrewrite(pynini.cross("zweitel", "halbe"), "[BOS]", "[EOS]", DAMO_SIGMA)
  28. integer = pynutil.delete("integer_part: \"") + change_one + pynutil.delete("\" ")
  29. numerator = pynutil.delete("numerator: \"") + change_one + pynutil.delete("\" ")
  30. denominator = (
  31. pynutil.delete("denominator: \"")
  32. + pynini.closure(DAMO_NOT_QUOTE)
  33. @ (
  34. pynini.cdrewrite(pynini.closure(ordinal.ordinal_stem, 0, 1), "", "[EOS]", DAMO_SIGMA)
  35. + pynutil.insert("tel")
  36. )
  37. @ change_numerator_two
  38. + pynutil.delete("\"")
  39. )
  40. integer += insert_space + pynini.closure(pynutil.insert("und ", weight=0.001), 0, 1)
  41. denominator_one_half = pynini.cdrewrite(pynini.cross("ein halbe", "ein halb"), "[BOS]", "[EOS]", DAMO_SIGMA)
  42. fraction_default = (numerator + insert_space + denominator) @ denominator_one_half
  43. self.graph = optional_sign + pynini.closure(integer, 0, 1) + fraction_default
  44. graph = self.graph + delete_preserve_order
  45. delete_tokens = self.delete_tokens(graph)
  46. self.fst = delete_tokens.optimize()