khmerphonemizer 0.0.3

A Free, Standalone and Open-Source Khmer Grapheme-to-Phonemes.

Khmer Phonemizer

A Free, Standalone and Open-Source Khmer Grapheme-to-Phonemes.

[Colab]

Installation

pip install khmerphonemizer

Usage

from khmerphonemizer import phonemize

text = "នៅលើលោកនេះមិនមានមនុស្សណាម្នាក់ចេះអស់ទេ"
result = phonemize(text)

print(result)

Output

(['នៅលើ',
  'លោក',
  'នេះ',
  'មិន',
  'មាន',
  'មនុស្ស',
  'ណា',
  'ម្នាក់',
  'ចេះ',
  'អស់',
  'ទេ'],
 [['n', 'ɨ', 'w', 'l', 'əː'],
  ['l', 'oː', 'k'],
  ['n', 'i', 'h'],
  ['m', 'ɨ', 'n'],
  ['m', 'i', 'ə', 'n'],
  ['m', 'ɔ', 'n', 'u', 'h'],
  ['n', 'aː'],
  ['m', 'n', 'ĕ', 'ə', 'ʔ'],
  ['c', 'e', 'h'],
  ['ʔ', 'ɑ', 'h'],
  ['t', 'eː']])

Check out the examples/ for more examples.

API

• phonemize Tokenize input text into words and phonemize each word and returns a tuple with tokens and phonemes.

  • input_str: str Text with multiple words.
  • beam: int = 500 number of beam search.
  • min_beam: int = 100: minimum number of beam search.
  • beam_score: float = 0.6 beam search score.
  • use_lexicon: bool = True Use lexicon dictionary for known words.

• phonemize_single Phonemize a single word.

  • word: str Text with single Khmer or English word only.
  • beam: int = 500 number of beam search.
  • min_beam: int = 100: minimum number of beam search.
  • beam_score: float = 0.6 beam search score.
  • use_lexicon: bool = True Use lexicon dictionary for known words.

License

MIT

---

References

Without these awesome projects from awesome people, this wouldn't be possible.

• Khmer Word Search: Challenges, Solutions, and Semantic-Aware Search (Rina Buoy and Nguonly Taing and Sovisal Chenda)
• CUNY-CL/wikipron (Kyle Gorman, Jackson Lee, and contributors, 2019)
• rhasspy/gruut (Michael Hansen et al., 2020)
• OpenFst (Kyle Gorman et al.)
• AdolfVonKleist/Phonetisaurus (Josef Novak et al., 2017)

Related

• khmercut
• khmernormalizer
• khmer-latin-name-transformer
• awesome-khmer-language