Spaces:
Running
Running
Commit
•
bae2ee4
1
Parent(s):
155155e
Upload symbols.py
Browse files- text/symbols.py +75 -0
text/symbols.py
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
'''
|
| 2 |
+
Defines the set of symbols used in text input to the model.
|
| 3 |
+
'''
|
| 4 |
+
|
| 5 |
+
'''# japanese_cleaners
|
| 6 |
+
_pad = '_'
|
| 7 |
+
_punctuation = ',.!?-'
|
| 8 |
+
_letters = 'AEINOQUabdefghijkmnoprstuvwyzʃʧ↓↑ '
|
| 9 |
+
'''
|
| 10 |
+
|
| 11 |
+
# japanese_cleaners2
|
| 12 |
+
_pad = '_'
|
| 13 |
+
_punctuation = ',.!?-~…'
|
| 14 |
+
_letters = 'AEINOQUabdefghijkmnoprstuvwyzʃʧʦ↓↑ '
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
'''# korean_cleaners
|
| 18 |
+
_pad = '_'
|
| 19 |
+
_punctuation = ',.!?…~'
|
| 20 |
+
_letters = 'ㄱㄴㄷㄹㅁㅂㅅㅇㅈㅊㅋㅌㅍㅎㄲㄸㅃㅆㅉㅏㅓㅗㅜㅡㅣㅐㅔ '
|
| 21 |
+
'''
|
| 22 |
+
|
| 23 |
+
'''# chinese_cleaners
|
| 24 |
+
_pad = '_'
|
| 25 |
+
_punctuation = '，。！？—…'
|
| 26 |
+
_letters = 'ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒㄓㄔㄕㄖㄗㄘㄙㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦㄧㄨㄩˉˊˇˋ˙ '
|
| 27 |
+
'''
|
| 28 |
+
|
| 29 |
+
'''# zh_ja_mixture_cleaners
|
| 30 |
+
_pad = '_'
|
| 31 |
+
_punctuation = ',.!?-~…'
|
| 32 |
+
_letters = 'AEINOQUabdefghijklmnoprstuvwyzʃʧʦɯɹəɥ⁼ʰ`→↓↑ '
|
| 33 |
+
'''
|
| 34 |
+
|
| 35 |
+
'''# sanskrit_cleaners
|
| 36 |
+
_pad = '_'
|
| 37 |
+
_punctuation = '।'
|
| 38 |
+
_letters = 'ँंःअआइईउऊऋएऐओऔकखगघङचछजझञटठडढणतथदधनपफबभमयरलळवशषसहऽािीुूृॄेैोौ्ॠॢ '
|
| 39 |
+
'''
|
| 40 |
+
|
| 41 |
+
'''# cjks_cleaners
|
| 42 |
+
_pad = '_'
|
| 43 |
+
_punctuation = ',.!?-~…'
|
| 44 |
+
_letters = 'NQabdefghijklmnopstuvwxyzʃʧʥʦɯɹəɥçɸɾβŋɦː⁼ʰ`^#*=→↓↑ '
|
| 45 |
+
'''
|
| 46 |
+
|
| 47 |
+
'''# thai_cleaners
|
| 48 |
+
_pad = '_'
|
| 49 |
+
_punctuation = '.!? '
|
| 50 |
+
_letters = 'กขฃคฆงจฉชซฌญฎฏฐฑฒณดตถทธนบปผฝพฟภมยรฤลวศษสหฬอฮฯะัาำิีึืุูเแโใไๅๆ็่้๊๋์'
|
| 51 |
+
'''
|
| 52 |
+
|
| 53 |
+
'''# cjke_cleaners2
|
| 54 |
+
_pad = '_'
|
| 55 |
+
_punctuation = ',.!?-~…'
|
| 56 |
+
_letters = 'NQabdefghijklmnopstuvwxyzɑæʃʑçɯɪɔɛɹðəɫɥɸʊɾʒθβŋɦ⁼ʰ`^#*=ˈˌ→↓↑ '
|
| 57 |
+
'''
|
| 58 |
+
|
| 59 |
+
'''# shanghainese_cleaners
|
| 60 |
+
_pad = '_'
|
| 61 |
+
_punctuation = ',.!?…'
|
| 62 |
+
_letters = 'abdfghiklmnopstuvyzøŋȵɑɔɕəɤɦɪɿʑʔʰ̩̃ᴀᴇ15678 '
|
| 63 |
+
'''
|
| 64 |
+
|
| 65 |
+
'''# chinese_dialect_cleaners
|
| 66 |
+
_pad = '_'
|
| 67 |
+
_punctuation = ',.!?~…─'
|
| 68 |
+
_letters = '#Nabdefghijklmnoprstuvwxyzæçøŋœȵɐɑɒɓɔɕɗɘəɚɛɜɣɤɦɪɭɯɵɷɸɻɾɿʂʅʊʋʌʏʑʔʦʮʰʷˀː˥˦˧˨˩̥̩̃̚αᴀᴇ↑↓∅ⱼ '
|
| 69 |
+
'''
|
| 70 |
+
|
| 71 |
+
# Export all symbols:
|
| 72 |
+
symbols = [_pad] + list(_punctuation) + list(_letters)
|
| 73 |
+
|
| 74 |
+
# Special symbol ids
|
| 75 |
+
SPACE_ID = symbols.index(" ")
|