30 lines
1.0 KiB
Python
30 lines
1.0 KiB
Python
""" from https://github.com/keithito/tacotron """
|
||
|
||
'''Defines the set of symbols used in text input to the model.
|
||
|
||
The default is a set of ASCII characters that works well for English
|
||
or text that has been run through Unidecode. For other data, you can
|
||
modify _characters. See TRAINING_DATA.md for details.
|
||
|
||
'''
|
||
|
||
_pad = '_'
|
||
_punctuation = '!\'(),.:;? '
|
||
_special = '-'
|
||
|
||
# Phonemes
|
||
_vowels = 'iyɨʉɯuɪʏʊeøɘəɵɤoɛœɜɞʌɔæɐaɶɑɒᵻ'
|
||
_non_pulmonic_consonants = 'ʘɓǀɗǃʄǂɠǁʛ'
|
||
_pulmonic_consonants = 'pbtdʈɖcɟkɡqɢʔɴŋɲɳnɱmʙrʀⱱɾɽɸβfvθðszʃʒʂʐçʝxɣχʁħʕhɦɬɮʋɹɻjɰlɭʎʟ'
|
||
_suprasegmentals = 'ˈˌːˑ'
|
||
_other_symbols = 'ʍwɥʜʢʡɕʑɺɧ'
|
||
_diacrilics = 'ɚ˞ɫ'
|
||
# some extra symbols that I found in from wiktionary ipa annotations
|
||
_extra_phons = ['g', 'ɝ', '̃', '̍', '̥', '̩', '̯', '͡']
|
||
|
||
phonemes = list(
|
||
_pad + _punctuation + _special + _vowels + _non_pulmonic_consonants
|
||
+ _pulmonic_consonants + _suprasegmentals + _other_symbols + _diacrilics) + _extra_phons
|
||
|
||
phonemes_set = set(phonemes)
|