File size: 4,858 Bytes
91394e0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7a23964
91394e0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
import json
import re
import warnings
from pathlib import Path

from kanjiconv import KanjiConv
from pypinyin import lazy_pinyin

from .resources.pinyin_dict import PINYIN_DICT

kanji_to_kana = KanjiConv()

yoon_map = {
    "ใ": "ใ‚",
    "ใƒ": "ใ„",
    "ใ…": "ใ†",
    "ใ‡": "ใˆ",
    "ใ‰": "ใŠ",
    "ใ‚ƒ": "ใ‚„",
    "ใ‚…": "ใ‚†",
    "ใ‚‡": "ใ‚ˆ",
    "ใ‚Ž": "ใ‚",
}

# ACE_phonemes
with open(Path(__file__).parent / "resources" / "all_plans.json", "r") as f:
    ace_phonemes_all_plans = json.load(f)
for plan in ace_phonemes_all_plans["plans"]:
    if plan["language"] == "zh":
        ace_phonemes_zh_plan = plan
        break


def preprocess_text(text: str, language: str) -> list[str]:
    text = text.replace(" ", "")
    if language == "mandarin":
        text_list = to_pinyin(text)
    elif language == "japanese":
        text_list = to_kana(text)
    else:
        raise ValueError(f"Other languages are not supported")
    return text_list


def to_pinyin(text: str) -> list[str]:
    pinyin_list = lazy_pinyin(text)
    text_list = []
    for text in pinyin_list:
        if text[0] == "S" or text[0] == "A" or text[0] == "-":
            sp_strs = re.findall(r"-|AP|SP", text)
            for phn in sp_strs:
                text_list.append(phn)
        else:
            text_list.append(text)
    return text_list


def replace_chouonpu(hiragana_text: str) -> str:
    """processใ€Œใƒผใ€since the previous packages didn't support"""
    vowels = {
        "ใ‚": "ใ‚",
        "ใ„": "ใ„",
        "ใ†": "ใ†",
        "ใˆ": "ใˆ",
        "ใŠ": "ใ†",
        "ใ‹": "ใ‚",
        "ใ": "ใ„",
        "ใ": "ใ†",
        "ใ‘": "ใˆ",
        "ใ“": "ใ†",
        "ใ•": "ใ‚",
        "ใ—": "ใ„",
        "ใ™": "ใ†",
        "ใ›": "ใˆ",
        "ใ": "ใ†",
        "ใŸ": "ใ‚",
        "ใก": "ใ„",
        "ใค": "ใ†",
        "ใฆ": "ใˆ",
        "ใจ": "ใ†",
        "ใช": "ใ‚",
        "ใซ": "ใ„",
        "ใฌ": "ใ†",
        "ใญ": "ใˆ",
        "ใฎ": "ใ†",
        "ใฏ": "ใ‚",
        "ใฒ": "ใ„",
        "ใต": "ใ†",
        "ใธ": "ใˆ",
        "ใป": "ใ†",
        "ใพ": "ใ‚",
        "ใฟ": "ใ„",
        "ใ‚€": "ใ†",
        "ใ‚": "ใˆ",
        "ใ‚‚": "ใ†",
        "ใ‚„": "ใ‚",
        "ใ‚†": "ใ†",
        "ใ‚ˆ": "ใ†",
        "ใ‚‰": "ใ‚",
        "ใ‚Š": "ใ„",
        "ใ‚‹": "ใ†",
        "ใ‚Œ": "ใˆ",
        "ใ‚": "ใ†",
        "ใ‚": "ใ‚",
        "ใ‚’": "ใ†",
    }
    new_text = []
    for i, char in enumerate(hiragana_text):
        if char == "ใƒผ" and i > 0:
            prev_char = new_text[-1]
            if prev_char in yoon_map:
                prev_char = yoon_map[prev_char]
            new_text.append(vowels.get(prev_char, prev_char))
        else:
            new_text.append(char)
    return "".join(new_text)


def to_kana(text: str) -> list[str]:
    hiragana_text = kanji_to_kana.to_hiragana(text.replace(" ", ""))
    hiragana_text_wl = replace_chouonpu(hiragana_text).split(" ")
    final_ls = []
    for subword in hiragana_text_wl:
        sl_prev = 0
        for i in range(len(subword) - 1):
            if sl_prev >= len(subword) - 1:
                break
            sl = sl_prev + 1
            if subword[sl] in yoon_map:
                final_ls.append(subword[sl_prev : sl + 1])
                sl_prev += 2
            else:
                final_ls.append(subword[sl_prev])
                sl_prev += 1
        final_ls.append(subword[sl_prev])
    return final_ls


def kana_to_phonemes_openjtalk(kana: str) -> list[str]:
    import pyopenjtalk

    with warnings.catch_warnings(record=True) as w:
        warnings.simplefilter("always")
        # add space between each character
        kana = " ".join(list(kana))
        # phones is a str object separated by space
        phones = pyopenjtalk.g2p(kana, kana=False)
        if len(w) > 0:
            for warning in w:
                if "No phoneme" in str(warning.message):
                    raise ValueError(f"No phoneme found for {kana}. {warning.message}")
    phones = phones.split(" ")
    return phones


def pinyin_to_phonemes_opencpop(pinyin: str) -> list[str]:
    pinyin = pinyin.lower()
    if pinyin in ace_phonemes_zh_plan["dict"]:
        phns = ace_phonemes_zh_plan["dict"][pinyin]
        return phns
    elif pinyin in ace_phonemes_zh_plan["syllable_alias"]:
        phns = ace_phonemes_zh_plan["dict"][
            ace_phonemes_zh_plan["syllable_alias"][pinyin]
        ]
        return phns
    else:
        raise ValueError(f"{pinyin} not registered in Opencpop phoneme dict")


def pinyin_to_phonemes_ace(pinyin: str) -> list[str]:
    pinyin = pinyin.lower()
    if pinyin in PINYIN_DICT:
        phns = PINYIN_DICT[pinyin]
        return phns
    else:
        raise ValueError(f"{pinyin} not registered in ACE phoneme dict")