Artrajz's picture
init
960cd20
raw
history blame
1.75 kB
from bert_vits2.text import chinese, japanese, english, cleaned_text_to_sequence, japanese_v111, chinese_v100, \
japanese_v200, english_v200, english_v230, chinese_v240, japanese_extra
language_module_map = {
'zh': chinese,
'ja': japanese,
'en': english,
'ja_v111': japanese_v111,
'zh_v100': chinese_v100,
'ja_v200': japanese_v200,
'en_v200': english_v200,
'en_v230': english_v230,
'zh_v240': chinese_v240,
'ja_extra': japanese_extra,
}
# _loaded_modules = {}
#
#
# def get_language_module(language):
# if language not in _loaded_modules:
# module_path = language_module_map.get(language)
# if not module_path:
# raise ValueError(f"Unsupported language: {language}")
#
# _loaded_modules[language] = importlib.import_module(module_path)
#
# return _loaded_modules[language]
def clean_text(text, language, tokenizer, pinyinPlus=None):
language_module = language_module_map[language]
norm_text = language_module.text_normalize(text)
phones, tones, word2ph = language_module.g2p(norm_text, tokenizer=tokenizer, pinyinPlus=pinyinPlus)
return norm_text, phones, tones, word2ph
# def clean_text_bert(text, language, tokenizer):
# language_module = language_module_map[language]
# norm_text = language_module.text_normalize(text)
# phones, tones, word2ph = language_module.g2p(norm_text, tokenizer)
# bert = language_module.get_bert_feature(norm_text, word2ph)
# return phones, tones, bert
def text_to_sequence(text, language, tokenizer):
norm_text, phones, tones, word2ph = clean_text(text, language, tokenizer)
return cleaned_text_to_sequence(phones, tones, language)
if __name__ == '__main__':
pass