File size: 5,015 Bytes
c5ed230
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
import re
from unidecode import unidecode
import pyopenjtalk


# Regular expression matching Japanese without punctuation marks:
_japanese_characters = re.compile(
    r'[A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]')

# Regular expression matching non-Japanese characters or punctuation marks:
_japanese_marks = re.compile(
    r'[^A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]')

# List of (symbol, Japanese) pairs for marks:
_symbols_to_japanese = [(re.compile('%s' % x[0]), x[1]) for x in [
    ('%', 'パーセント')
]]

# List of (romaji, ipa) pairs for marks:
_romaji_to_ipa = [(re.compile('%s' % x[0]), x[1]) for x in [
    ('ts', 'ʦ'),
    ('u', 'ɯ'),
    ('j', 'ʥ'),
    ('y', 'j'),
    ('ni', 'n^i'),
    ('nj', 'n^'),
    ('hi', 'çi'),
    ('hj', 'ç'),
    ('f', 'ɸ'),
    ('I', 'i*'),
    ('U', 'ɯ*'),
    ('r', 'ɾ')
]]

# List of (romaji, ipa2) pairs for marks:
_romaji_to_ipa2 = [(re.compile('%s' % x[0]), x[1]) for x in [
    ('u', 'ɯ'),
    ('ʧ', 'tʃ'),
    ('j', 'dʑ'),
    ('y', 'j'),
    ('ni', 'n^i'),
    ('nj', 'n^'),
    ('hi', 'çi'),
    ('hj', 'ç'),
    ('f', 'ɸ'),
    ('I', 'i*'),
    ('U', 'ɯ*'),
    ('r', 'ɾ')
]]

# List of (consonant, sokuon) pairs:
_real_sokuon = [(re.compile('%s' % x[0]), x[1]) for x in [
    (r'Q([↑↓]*[kg])', r'k#\1'),
    (r'Q([↑↓]*[tdjʧ])', r't#\1'),
    (r'Q([↑↓]*[sʃ])', r's\1'),
    (r'Q([↑↓]*[pb])', r'p#\1')
]]

# List of (consonant, hatsuon) pairs:
_real_hatsuon = [(re.compile('%s' % x[0]), x[1]) for x in [
    (r'N([↑↓]*[pbm])', r'm\1'),
    (r'N([↑↓]*[ʧʥj])', r'n^\1'),
    (r'N([↑↓]*[tdn])', r'n\1'),
    (r'N([↑↓]*[kg])', r'ŋ\1')
]]


def symbols_to_japanese(text):
    for regex, replacement in _symbols_to_japanese:
        text = re.sub(regex, replacement, text)
    return text


def japanese_to_romaji_with_accent(text):
    '''Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html'''
    text = symbols_to_japanese(text)
    sentences = re.split(_japanese_marks, text)
    marks = re.findall(_japanese_marks, text)
    text = ''
    for i, sentence in enumerate(sentences):
        if re.match(_japanese_characters, sentence):
            if text != '':
                text += ' '
            labels = pyopenjtalk.extract_fullcontext(sentence)
            for n, label in enumerate(labels):
                phoneme = re.search(r'\-([^\+]*)\+', label).group(1)
                if phoneme not in ['sil', 'pau']:
                    text += phoneme.replace('ch', 'ʧ').replace('sh',
                                                               'ʃ').replace('cl', 'Q')
                else:
                    continue
                # n_moras = int(re.search(r'/F:(\d+)_', label).group(1))
                a1 = int(re.search(r"/A:(\-?[0-9]+)\+", label).group(1))
                a2 = int(re.search(r"\+(\d+)\+", label).group(1))
                a3 = int(re.search(r"\+(\d+)/", label).group(1))
                if re.search(r'\-([^\+]*)\+', labels[n + 1]).group(1) in ['sil', 'pau']:
                    a2_next = -1
                else:
                    a2_next = int(
                        re.search(r"\+(\d+)\+", labels[n + 1]).group(1))
                # Accent phrase boundary
                if a3 == 1 and a2_next == 1:
                    text += ' '
                # Falling
                elif a1 == 0 and a2_next == a2 + 1:
                    text += '↓'
                # Rising
                elif a2 == 1 and a2_next == 2:
                    text += '↑'
        if i < len(marks):
            text += unidecode(marks[i]).replace(' ', '')
    return text


def get_real_sokuon(text):
    for regex, replacement in _real_sokuon:
        text = re.sub(regex, replacement, text)
    return text


def get_real_hatsuon(text):
    for regex, replacement in _real_hatsuon:
        text = re.sub(regex, replacement, text)
    return text


def japanese_to_ipa(text):
    text = japanese_to_romaji_with_accent(text).replace('...', '…')
    text = re.sub(
        r'([aiueo])\1+', lambda x: x.group(0)[0]+'ː'*(len(x.group(0))-1), text)
    text = get_real_sokuon(text)
    text = get_real_hatsuon(text)
    for regex, replacement in _romaji_to_ipa:
        text = re.sub(regex, replacement, text)
    return text


def japanese_to_ipa2(text):
    text = japanese_to_romaji_with_accent(text).replace('...', '…')
    text = get_real_sokuon(text)
    text = get_real_hatsuon(text)
    for regex, replacement in _romaji_to_ipa2:
        text = re.sub(regex, replacement, text)
    return text


def japanese_to_ipa3(text):
    text = japanese_to_ipa2(text).replace('n^', 'ȵ').replace(
        'ʃ', 'ɕ').replace('*', '\u0325').replace('#', '\u031a')
    text = re.sub(
        r'([aiɯeo])\1+', lambda x: x.group(0)[0]+'ː'*(len(x.group(0))-1), text)
    text = re.sub(r'((?:^|\s)(?:ts|tɕ|[kpt]))', r'\1ʰ', text)
    return text