def __call__(self, text) -> List[str]:
if self.g2p is None:
import g2pk
self.g2p = g2pk.G2p()
phones = list(
self.g2p(
text,
descriptive=self.descritive,
group_vowels=self.group_vowels,
to_syl=self.to_syl,
))
if self.no_space:
# remove space which represents word serapater
phones = list(filter(lambda s: s != " ", phones))
return phones
import re
from unicodedata import normalize
sys.path.append("/content/src/glow-tts")
from text import symbols as glowtts_symbols
sys.path.remove("/content/src/glow-tts")
sys.path.append("/content/src/g2pK")
import g2pk
sys.path.remove("/content/src/g2pK")
symbols = set(glowtts_symbols)
g2p = g2pk.G2p()
def normalize_text(text):
text = simple_replace(text)
text = g2p.idioms(text)
text = g2pk.english.convert_eng(text, g2p.cmu)
text = g2pk.utils.annotate(text, g2p.mecab)
text = g2pk.numerals.convert_num(text)
text = re.sub("/[PJEB]", "", text)
text = eng_cap(text)
text = normalize("NFD", text)
for pos, char in enumerate(text):
if char not in symbols:
text = text[:pos] + " " + text[pos + 1:]